Jim's AMA (11 January 2023)
An edited transcript
In perhaps the most seismic social media event to occur since Kim Kardashian broke the internet, Jim appeared on a Twitter Space hosted by our colleague Matt Ryan on 11 January 2023 to answer listeners’ questions about O’Shaughnessy Ventures and its four verticals.
For those of you who weren’t able to join, we have shared a transcript of the conversation below (please note that the transcript has been heavily edited for clarity and brevity).
A full, unedited recording of the discussion is available here.
Matt Ryan:
Welcome, everybody. This is the first Twitter Space that Jim has done since he launched O'Shaughnessy Ventures. We're going to do this in an AMA format, so I'm sure that Jim will have probably several remarks to open. From there, we will go ahead and slide on into the questions. Just request to speak. I will pick whoever I like based on completely arbitrary criteria, or how lazy I am to not scroll down.
We'll bring you in as a speaker, ask your question, and then we'll pop you back out. Jim will give probably an entirely too long answer. Fun item to note here. The last time Jim had this many people listening to him speak this late in the evening, I believe he overthrew several democratically elected South American governments. So, this is a little bit of a change of pace for him in that regard. Without any further ado, Jim, why don't you go ahead and take it away?
Jim O'Shaughnessy:
Thanks, Matt. Thanks, everyone, for showing up at eight o'clock on a weekday. I appreciate it. I really don't have a lot to open up with. I really wanted this to be as open and give-and-take as possible. A lot of people have DM-ed, or emailed, or whatever, me questions, and I thought this would just be a better opportunity to just do it live, as in the Bill O'Reilly, "Let's do it live! Do it live! Fuck it!" There is swearing allowable. So, if you do wish to curse, go right ahead. But honestly and sincerely, thank you all for coming. I'm super-excited about the new company. With that, Matt, pick somebody.
Adam:
First and foremost is, can you help me better understand what your new O'Shaughnessy Ventures is?
Jim O'Shaughnessy:
Sure. O'Shaughnessy Ventures is essentially the umbrella organization that houses four verticals, each of which I have been really, keenly interested in starting a company in and/or contributing in a field. But that's rather difficult to do when you are the head of a traditional long-only asset management firm. Essentially, if you go to the website, you'll see the verticals, which are basically the idea of the fellowships, the idea behind venture capital, which we call Adventure Capital, which was its original name.
I've always loved films, one of my hobbies was to write treatments for films. So, I've got a bunch of notebooks filled with ideas for them. Also, got tired of the fact that Hollywood stopped making movies like Rudy, that are inspirational and uplifting in terms of showing that persistence and dogged determination can get you somewhere.
Then, Infinite Media. Everyone says we're oversaturated in media. I think that that is wrong. I think that you're going to see a whole new slew of both podcasts, YouTube channels, Substacks, et cetera, coming online that are going to be of really outstanding quality. I think that there are a lot of business opportunities there, as well as just support opportunities.
So, given the fact that I retired at the end of the year from O'Shaughnessy Asset Management... Still very, very fond of everyone there. I think it's a fantastic team, and I think they will crush the custom indexing market, which is a new category in asset management. I'm off on this adventure. I guess the one I'm best known for, obviously, would be the investing side, but that is really just one of the verticals.
There, we're going to do it a lot differently than we did at OSAM, in that it's going to be mostly private, not public, mostly seed to Series A, not well known existing companies. But that is something that I have been doing since early 2000 or late 2006, early 2007. So, we've gotten a lot of experience, both good and bad, there. I've just found it to be a really fascinating market, and also a fascinating time. I think we are going through a Cambrian-like explosion of innovation, creativity, and many of that will find its way into a pitch deck.
O'Shaughnessy Ventures is each of those verticals, neither more important than the other. Then of course the final isn't really a vertical, but it's something that I've wanted to do for a long time. This is the idea of the awarding of fellowships to creative, innovative people. I'll take you, at some point in this Spaces, through just some of the early applications. I am blown away by the amount of talent in the world.
As the world has shrunk, geography, space, time have all collapsed, we really have no reason not to be able to find incredible talent and brilliance in the world. When we look at history, how many geniuses were born, live and died, and nobody ever heard of them or of their ideas. Many times, they didn't even know they were gifted. It was a very static-type society, which leads to not great things, in my opinion. I think we're going into a very dynamic one where all of these people can be found, and we're going to find them and fund them.
Adam:
I'm going to do a follow-up. Would that be all right?
Jim O'Shaughnessy:
Of course.
Adam:
Okay. That's awesome.
Matt Ryan:
I will allow it.
Adam:
Oh, that's awesome. That's very thoughtful of you. Would you mind, I see that it says, O'Shaughnessy, that you are the host. I'm referring there to the person who's actually, I suppose, conducting this Space. Would you mind identifying yourself? Would that be all right?
Matt Ryan:
Yes, absolutely. My name is Matthew Ryan and I handle social media, performance analytics, and some other odds and ends there for Jim, for this venture.
Adam:
Allow me just one more observation, and then I really do have a question. Another question is, that's very interesting, your own interest in screen treatments and movies. It's actually something, an interest, I've acquired over the last several years, and I've read a lot of books on it. I've read books called Save the Cat. I've read so many books on the subject of how to tell a story, how a story is crafted both in the novel, the literary form, and then also in the movie form, and the pacing and the different kind of devices that there is. Really is very, very interesting.
Jim O'Shaughnessy:
All right. Yeah.
Adam:
Yeah, that was my observation part. Then, the question part of it is your involvement with Stability AI. Then, just adding a little bit more force into my question is, now that we have so much interest in the new OpenAI, and that it's getting some valuation now, which is something it didn't have before, and that people are racing for leadership within that, and there's really not any fail-safe, the public has not yet decided which one is their favorite, could you help me better understand Stability AI, and then your own participation in it?
Jim O'Shaughnessy:
Sure. AI, artificial intelligence, machine learning is at an inflection point, and you're seeing the fruits of actually something we did, and attracted me to Stability AI, when our founder, Emad Mostaque, released an entire AI generative model with the weights. Previously, models had been released on a more limited basis, making them very difficult for developers to use to develop new products, new routines, et cetera.
By doing that, we wanted to demonstrate that open is better than closed, because it allows for a multiplicity of creators and innovators to use the model to develop APIs, apps, et cetera, because without the weights, you can't do any of the fine-tuning. That is not allowed in a closed AI environment. So, it's open source versus closed source. That's another biggie as far as I'm concerned, given the fact that in open source, for example, we have 50,000 coders working on our various multi-modal model systems, and they come from a variety of backgrounds with a variety of skills.
So, very much like Linux, that type of system tends to make greater advances in a shorter period of time than those where the coding encoders are not allowed. There are also different types of models when you have, for example, Chat or GPT Chat is a language model known as a large language model. Those are usually simply trained to predict the next character of a vast dataset of text from the internet, is what it's usually trained on. But they can be trained on private data sets as well, literature, et cetera.
Those, generally speaking, do very well for chatbots. You're seeing all of the excitement about GPT Chat, excellent model for that type of use case. They're also great at code generation, assisting software developers, et cetera. Foundational models are those that are trained on... and both of these, by the way, are foundational models, but stable diffusion models, image models, are essentially trained not only on text but also on a vast quantity of images as well.
So, the value in a generative AI model is such that the model is trained on data that gives a probability distribution on the underlying data. For example, if you're trying to predict the next character, the given tech sequence, that tends to be a little bit easier than if you are trying to form based on a probability distribution, a model that allows it to generate new data that looks somewhat like the data it was trained on, but is not exactly the same as the training data.
It seems to be a subtle difference, but the downstream use cases for these types of models, I think, can be very different. I think that what you're going to see is foundational models as a service where you go into anything from game makers to filmmakers, to think of virtually any profession, and train it on their private data sets so that it becomes unique to them. There would be, for example, an atom model if you wanted one, and hired Stability AI to train it for your specific uses.
On a broader scale, I think that the reason that I was so taken by Stability AI is its commitment to open source. I believe quite passionately that we really want as much diversity of input from the various creatives and innovators working on these models. Cognitive diversity is what's most important. I think it's the height of arrogance to assume that me, or a small group of people that I might put together, would know every conceivable use case for a model.
I think we proved that rather definitively when we released the model back in August, and all of a sudden had 200 various innovations, APIs, apps, et cetera, that many of us and some of the people that I was listening to, or geniuses at this, they openly said, "Gee, I never thought of that use case."
So, I think having the open gives you several competitive advantages, but it also gives advantages to the end user and to the developers, in that it gives them a fantastic tool and model on which they can build really, really cool apps. I think you're seeing a ton of them being released as we speak, and I think you're going to see tons more coming down the pipe. Then finally, just a philosophical difference for me was, I have grandchildren and they're going to grow up in this world. I want them to grow up in a world where it is not a giant panopticon controlled by a few telling them what they can and can't do with this incredibly powerful technology.
Yes, there are dangers. I think they need to be addressed. We are trying to do that as best we can at Stability AI, and we have many initiatives to address how to handle those things. But at the end of the day, all tools can be used for both good purposes and bad purposes. Fire was really good for us and our developing brains, but it also could be very destructive. Therefore, we have fire departments, fire warnings, fire alarms, et cetera, and we are taking that very, very seriously as well.
On balance, I think that the foundational generative models have more downstream use cases that can be trained on a variety of different use cases for different industries, and therefore think that in the long run, Stability AI will be the company that is that open AI company, and kind of a Switzerland. We don't want to be captured by any particular large corporation, and we want to be able to service as many clients without having conflicts of, "We want to develop our own version of..." say, Canva or Adobe. We don't.
We don't want to put Shutterstock or Getty Photos out of business. We want to make them better. So, you'll see us as trying to maintain as much neutrality as possible, and not trying to develop products, or apps, or APIs that would compete directly with the clients that we hope that we will be working with as we proceed.
Adam:
Yeah, yeah. Thank you very much. I'm very interested in what you're doing with your fellowship, and the candidates you're seeing, and the direction of the fellowship.
Jim O'Shaughnessy:
Thank you, Adam. I'm quite passionate about the fellowship and the terms, more specifically, of the fellowship. There are no terms. When we select you as a fellow, you don't have to sign anything. You don't have to promise to give us your IP. You don't even have to promise to work with us, should you decide that you want to start a company based on the work that you did when you had the luxury of working full-time on your passion project.
Really, what we want to do and demonstrate is that, number one, we are now at a period in history where people don't have to be lost because of where they live, or what color they are, or what sex they are, or what particular creed, or anything else. They can in fact be judged on the quality of their ideas. We want to both let them do that, but we also are doing this to highlight the fact that there are so many incredibly brilliant people in the world.
Now, I get chided by many when I say "young people", and it doesn't mean I don't think that older people like me and you, Adam, have bad ideas. I think we have great ideas, but generally younger people often are in positions where their need to make a living gets in the way of their desire to pursue a passionate idea or life. So, the fellowship was also conceived of as something that would allow them the ability to do that.
Also, we want to highlight, there's so much doom and gloom in the world today, and I want to counterbalance that, even just the tiny little bit that I can, by showing you the number of incredibly brilliant people that are around. For example, in the first 10 days, we've gotten over 400 applications, and here are some of the things that people are applying for, that they want to study under the fellowship.
We have an applicant from South Africa who wants to research mammals' experience of consciousness, and create a blueprint for machines to experience it. We have a rocket engineer who wants to build an educational initiative to create the world's best rocket engineers.
We have another applicant who wants to build 3D human tissues for faster drug development, lab-grown human tissue, so that we can recreate... after doing a DNA scan of you, Adam, or me, Jim... our DNA, our essence, in a lab. That they can do things to that they would never do to a living human being. Now, that also ties in with AI, because we can do a lot of these experiments in what they call "in silico", which means on the computer, but this also is an interesting one. Yet another wants to build the first venture capital fund in Somalia. And so, as you look at these things, they just really jump off the page, at least for me they do. A final one here, we have an applicant who's stationed in Ukraine, and he or she, I don't know if it's a he or she, wants to focus and spend their fellowship trying to develop a process for evacuating injured people, and build an app that helps people operate their way through a conflict zone. So, these are just the beginning applications. We anticipate that we will receive many, many more, and the really hard part, I think, for us and the team that is evaluating all of these applications...
Boy, if we had our way, we give a fellowship to everybody. But we don't. We unfortunately don't have a money tree that we can just pull the money off of. But I think already, in these early days, we are seeing just such a tremendous amount of incredibly great ideas that deserve to be heard, that deserve to be supported and funded. And so, that is why I'm so excited about the fellowship. And, of course, we're in business to make money, so I'm sure that we will end up funding some of the more entrepreneurial fellows, but they're not required to come with us. They could take a bid from Sequoia or someone else if they wanted to. We just wanted to unlock the talent, demonstrate it's there, and show how many great ideas actually are existing in the world. And I think just from these very, very early days that we're going to see that in a major way. Matt, you want to...?
Matt Ryan:
Absolutely. Adam, thank you so much for the wonderful question. Next up, we're going to have Nat. Give me just a second here as we bring Nat in. All right, Nat. You are and should be live. Go ahead and fire away.
Nat:
Hey Jim. Yeah, super exciting. I'm very happy you are doing this. It'll be fun to see what kind of ideas surface. I have a couple more detailed questions just on the fellowships. It sounds like it's extremely open-ended, very generous. But what would a day-to-day kind of thing look like? Would there be a reporting structure? Would there be check-ins with you or other members of OSV? What are you thinking in that sense?
Jim O'Shaughnessy:
So, yeah. There will be check-ins with the relevant OSV team member who probably has the better domain knowledge. And they're not going to be bureaucratic in any sense of the word. We are here to share the resources and our network. Luckily, I have a pretty good network that spans quantum physicists all the way to venture capitalists and traditional founders and entrepreneurs who've run businesses and know the sometimes unpleasant details of what it takes to actually run a business. So, the way we're fashioning it right now, Nat, is we're here to help. And obviously, we think that having our fellows check in with us and have conversations with us will help them too, because hopefully we'll be able to ask reasonably intelligent questions that maybe the fellow hadn't thought of, and iterate that way. It very much is going to be a discussion more than a lecture. We'll also, if a fellow, we think, is really going the wrong way, we'll tell them that, but we don't have any veto. We're not retaining a veto to ourselves.
I'm sure if there was an extreme case where the fellow was literally doing nothing and just having rave parties every night, where we could rescind a fellowship if they're not doing what they told us they wanted to do. But other than that, it's very much in my spirit of let creative people be creative. And they, for the most part, are not terribly good rule followers. One of the things on our application is the last one, where we ask if you're working or not, and the last one is, "I am unemployable." I make that joke about myself all the time. But unemployable people are interesting people, and we are very open to people who don't fit the standard neurotypical mold. In fact, we think most of the really interesting stuff is going to come from the tales. And so, yes for support, yes give access to our networks, our people, our relationships. No, we're not going to be in any way telling them what they can and can't do. That would, in my mind at least, defeat the purpose of what we're trying to achieve here.
Nat:
Got you. Well, my wife and I are definitely on our way to being unemployable, so that's good. I'm doing my best. I'm trying. I'm trying, Jim. Okay, so, the final follow-up then is just, at the end of that year, you have all these great ideas, all these people coming to you with these really cool concepts, but what does a success story look like to you, at the end of that year?
Jim O'Shaughnessy:
So, it would depend on what the fellow received the fellowship for. So, if it was, let's take the applicant who wants to build the first VC fund in Somalia. Let's say he or she won a fellowship. Success obviously would look like us seeding their VC and giving them support and structure around how you do that, what you need to have in place to run a venture capital firm successfully, et cetera. In the case of an artistic fellow, let's say we get somebody who wants to study and use new materials for creating sculptures. Success would look like, at the end of that, that he or she would have successfully found materials that hadn't to date been used as the base material for a sculpture and is making beautiful sculptures with them, and that then we could introduce them to one of the many art galleries that we have relationships with. So, success looks like them using the time well, executing against their vision, and then, at the end of the fellowship, creating a situation where we can either fund them if they are a enterprise or it's an entrepreneur, we can give introductions to galleries in the case of an artist. And then, let's just take a scientist, let's say, because we also have one that, he wants to open source quantum computing. And let's say he got the fellowship and was successful in uncovering either better groups or better participants to make that a success, then we could introduce them to various think tanks like the Santa Fe Institute or universities that are working on that. So, we have the luxury of being able to make some pretty good introductions where it isn't just a matter of us seeding a company's first seed round, for example.
Matt Ryan:
Fantastic. Thank you so much, Nat. All right. Next up is going to be We Live to Serve. So, as soon as that pops in, there you go. We Live to Serve. You are alive, so go ahead whenever you're ready.
Keith:
I wanted to ask you. So, you know my whole thing is dad life, and from the beginning I've been trying to give my kids creative ways to learn about, especially geometry, but just concepts of what numbers are and how those relationships work. And so, what's crazy to me is that this mathematics that goes back a long time, in particular abstract algebra and number theory, that not only does it come into play in physics, relativity and quantum physics in particular, but that same sort of matrix manipulation and the Lie algebra rules and symmetries and whatnot is very much in play with AI and machine learning, because of the matrix structure of all this.
And so, that's a long-winded way of asking you, combining your passions for your grandkids and the great reshuffle. If you were to give Dutch uncle advice to, let's say three ages, like 25 years old, 15 years old, and five years old, obviously understanding that creativity and love of learning is table stakes and it's a growth mindset, and we get them out of this Taylorian bullshit of nine to five, what are the things that you would want your grandkids to be learning to just be... Not everyone's going to be a mathematician. And I think one of the byproduct great results of this revolution is that it's actually brought coding more to the level of an individual, just somewhat savvy person, more than ever before. But, long-winded way, what do you think are the kind of table stake skill sets for young people from a technical, and to just be ready for this new world?
Jim O'Shaughnessy:
Yeah, that's a great question, Keith. And I know how involved you are with your kids, so kudos on that. I'm going to give a slightly broader answer, because I think that you've got to get your first principles in place to make those other things happen. So, I think that we have education wrong, especially for younger people. It is an antiquated system that essentially was set up during the industrial era, where the industrial barons of the day wanted people who were literate. So, they needed to be able to read. They wanted them to be able to write and do basic calculations. But they also wanted them to be compliant, and they wanted them to be able to do a repetitive task for eight to 10 hours a day in the same room, and all the other wicked Taylorisms that you know about and I know about. And that world vanished. And yet, we continue with that sort of creativity-crushing educational method of teaching.
And so, I think the first thing that I would wish for my grandchildren is that they grow up in an environment where they learn how to think, not what to think. And that sounds like it might be a minor difference, but it's a major, major difference. Most of our so-called learning still is on teaching to a test, to a fact. "What happened in 1066?" Okay, so, I know it's the Battle of Hastings, but do you need to know that if you have access to an AI on your cell phone, that can train itself on you and your peculiar way of learning, because everyone does have a different learning style?
So, I had lunch, I just got back from Miami and I had lunch with one of the people from Synthesis School. And I think they're doing it the right way. They're teaching kids how to think. They're putting kids from a variety of cultures and countries together in the same Zoom. I don't think they use Zoom, they use a different tech. And so, they're learning about other cultures and how other cultures think about things. We do not have a monopoly on science or innovation. There's a great book called ‘The WEIRDest People in the World’ that outlines how many of the tests of attitudinal tests and learning and things like that are heavily skewed to the United States and the UK, and to a lesser extent other English-speaking countries. And we are very, very different when you look at cross-sectional analysis of populations. Western countries, and thus the title, are weird, and the rest of the world doesn't really share many of the traits that, for a particular material type of success, we have excelled at in the West.
So, I would begin by saying, first, teach them how to think, not what to think. They will have access, especially your five-year-old, I think, will have access to an AI-enabled tablet or phone that trains itself on your child and is able to iterate very, very quickly. And almost, it seems magical, that they start asking questions of the child in a vernacular or a structure that the child immediately responds to. That goes back to the days of neurolinguistic programming and everyone having a different communication style. Most of us are visual communicators, but many of us are auditory and/or kinesthetic. There's a variety of sub-modalities. The AI can train itself on the child's peculiar learning style and help them learn. And then I would say, you said table stakes. So, I think table stakes today is going to be, you're going to have to learn how to unlearn. And that sounds weird.
Keith:
Right. Table stakes are forgetting the table stakes.
Jim O'Shaughnessy:
Yeah, right. And I've had to unlearn a lot, and learn new things, and I'm going to have to unlearn things that I just learned because they'll get displaced by new innovations, new discoveries, new theses, new hypotheses that are going to be at odds with what I think right now. And if you think historically, that shouldn't surprise any of us. If you look back at the smartest human being, maybe 700 years ago, everything they thought they absolutely knew for certain was wrong. Not everything, but almost everything. And so, I think that, I don't know if it's an imprinting deal, the idea of how character and personality develops, or it's a genetic deal. But it doesn't have to be.
I think, just go on YouTube. I was a YouTube babe. I didn't use YouTube. And some of my teammates were said, "You have to do a deep dive on YouTube." And so, I did, and my God, it's amazing. You don't know how to tie your tie. Well, you can learn from YouTube. You don't know how to sand a floor. You don't know how to build a cabin. I was just talking to Meb on his podcast, and he said he was watching YouTube with his kids about a guy, I think in Finland, building a cabin. And it was time lapsed over three years, and it had 20 million views. So, I think that with the cornucopia of these new learning agents, systems, platforms, et cetera, that if you can get a child in the mode of not beating their curiosity and creativity out of them, which I think our current system does, it doesn't like things that don't fit into the right hole or square, and it tries to beat them into submitting to that particular thing that they want as an outcome.
I think that's got to go. I think it will go. And if we get them loving figuring stuff out, loving learning, being willing to understand that they're going to need to unlearn, then they'll be able to find ample resources. You asked specifically about geometry and whatnot. We are seeing courses, even one of our applicants here, actually has a new methodology for teaching mathematics and she wants to make it go more broadly. And so, those things are going to be ubiquitous and plentiful online, and most of them are going to be free. Digital things-
Keith:
Yeah, it's pretty mind-blowing.
Jim O'Shaughnessy:
Yeah. So, digital things with no additional cost to creating them end up going to be basically free. Now, what isn't free? Trust isn't free, integrity isn't free, learning how to learn isn't free. You've got to have a disposition for that. But there will be, in the next five years... I'll give you an example, both from here in the States and then something Stability AI is involved in. Here in the United States, I learned from my friend Anna at the Synthesis School that they're using AI-aided educational tools, and they're getting remarkable results by moving kids who are in the bottom decile of a particular category, I think math was the one that they were looking at the closest. But, in six to eight weeks, moving them up to the top decile, because the methodology of teaching is so incredibly different and can be customized to the child.
Keith:
Speaking of methodology, can I ask you a quick question?
Jim O'Shaughnessy:
Sure.
Keith:
So, as the ultimate quant, as this encyclopedic resource for historical data, how does that quant feel in a world where the analytical tools going forward aren't necessarily deterministic anymore?
Jim O'Shaughnessy:
I love it. In fact, I'm an unusual quant in that, obviously I prefer evidence and data to lack of evidence and no data. But given the new way of knowledge creation, it simply wasn't possible, because humans, as smart as we are and as smart as we collectively are as the human colossus, we simply cannot analyze an array of a hundred million variables in seconds, which allows us to see into latent spaces which were opaque to us in the past.
Keith:
That's exactly where I was going, I was thinking of Jeremiah's company, Prefect, in the sense that, at some point, much you talk about journaling so that you can be an inquisitor to an unreliable narrator, these models themselves, their weights, you have to have a way to preserve some of that kind of lab data, to know where they were at when they made those decisions over time. And so, to me, it combines the need for a rigorous approach to documentation to a certain point of view, but, again, you've got to find a way with these models to look back at the journal entry and just... You don't have to, but it's just interesting to think about how the underlying analytic engine can actually shift over time.
Jim O'Shaughnessy:
Yeah. And it will shift over time. It will get smarter and smarter and smarter. And speed is no longer, GPU access used to be incredibly expensive. It's still expensive, but it's not nearly as scarce as it was. What scarcer in AI is talent. David Ha, who at Stability AI, he left Google Brain to come and work for us at Stability AI. I asked him, "So, David, how many world class AI minds are out there?" And his answer shocked me. You want to take a guess?
Keith:
I mean, I'll say 18.
Jim O'Shaughnessy:
Yeah, well, I guessed high. I thought a lot more, given the population of the world. A hundred. There's a hundred world class talented people in AI. Now, that will change, obviously, because Stability AI is funding scholarships for PhDs in AI. Other firms will do it. We'll probably end up giving a fellowship at some point to somebody who wants to do it. But it is scaling really rapidly, and as it does so, one of the things that excites me is it will do... So, I'm, as you know, a huge fan of Claude Shannon's information theory.
Keith:
Same.
Jim O'Shaughnessy:
And one of the things that AI is going to be able to do, and I credit Liberty with this idea, it's not mine. But is, we should train an AI to simply generate scientific null sets, and then electronically publish them to a database that all scientists can have access to. Why? Because human beings don't want to work two years and ask for a grant on something they believe is going to lead to a null set. And yet, the data available, the information, rather, available in null sets, is huge.
It's like, to make it more prosaic, the Sherlock Holmes story where he knew that the dog knew the intruder because he didn't bark. It's via negativa. And there is so much information in those null sets that we don't really have access to, because human OS doesn't want to work on a project where you're not going to get published, you're not going to get accolades from your fellows. So, that's a great use case, where we can have an AI create a million, 10 million, 20 million null sets that scientists can draw upon. So, I definitely think, I'm quite comfortable with the new way of doing it, because we didn't have this capability up until just very recently.
Keith:
I agree, and I think we can build the intuition for it. I know you've read ‘Flatland’ by-
Jim O'Shaughnessy:
Yes, yeah.
Keith:
Edwin Abbott. But I will always remember where I was when I read that image of the sphere coming up through ‘Flatland’, and how it-
Jim O'Shaughnessy:
Brilliant.
Speaker 2:
Swelled on the way through.
Jim O'Shaughnessy:
Brilliant.
Keith:
And now, for the last, and again, I was always doing this from the point of view of the standard model of physics, and SO(3) groups and Lie algebras and fiber bundles. That was all a physics thing. But it's very much the case that a lot of this mathematics, you can get a better intuition for it. Think about a higher order sphere. Now, it's hard to picture what a seven-dimensional sphere looks like, but it's not hard to picture a dataset where, basically all a sphere means is that the entire surface of the sphere is a radius away. So, that's literally all it means. So, that idea of a circle symmetry, of a surface or a space of information where everything is equidistant to a central point, that is a very straightforward symmetry. And that's why, in higher order of dimensions, spheres go down to zero volume, because by definition they're compactifiers, whereas the square symmetry has a very different role. It's literally building out volumetric space for linear sorting.
Jim O'Shaughnessy:
And for those who don't want to read a more technical book, or want to read a less technical book, I would recommend the book ‘Unflattening’. That is a graphic novel that introduces image with words. It's going to be a new way of communicating, and I think it's going to be a superior way. And we're going to have to learn it. And a lot of people will, at first, resist it. And that's something we also have to deal with, but yeah. So, that's what I would say. So, Matt, let's pick another one, and Keith, good to speak to you.
Matt Ryan:
And Jim, just to tack in there, I'm honestly personally shocked that you don't have an eidetic memory to remember all 400 plus of those that came in, but there was actually several of those fellowship applications that dealt with new ways to teach math, especially to children, and especially concepts that we used to consider much higher level, to be taught later in life, and making them accessible at a young age in the way that we do it. And then, I guess another fun one to point out, there was a couple in there regarding programming, coding, some with a little towards AI targeting younger individuals as well, sub-18, just to circle back to that point that we were talking about there. So, some really, really truly great stuff that's coming in that vein when we're talking about children and education and table stakes. So, Tony, you're going to be the next one up here. So, whenever you are ready, slash whenever Twitter decides you're going to be ready, go ahead and fire away.
Tony:
Oh, thank you. Apologies. So, I have a bit of a logistics question about applying, and about how the fellowship would work. So, one of the big challenges that I've found is that, I've spent the last 10 years working in conventional and the last five years working in renewable energy. And in that five years, I have dedicated the last five years to human progress studies and communications and research. And I actually discovered this talk through Mateus from Warp News, who's an acquaintance.
Jim O'Shaughnessy:
Oh, terrific.
Tony:
Yeah. And so, I think what you're doing is brilliant. And one of the big challenges is, if you're like myself and you're mature in your career, I've spent the five years looking at other peers in my progress studies field and thinking, "I would just love to be able to unplug from work and instead of doing 12 hours a day of erecting renewable energy, which is amazing, do 12 hours a day of just smashing out amazing progress studies projects." But the ROI on that, as a handover to say a potential funder, can sometimes be tenuous or not so obvious. And so my question really is, what's your view on funding things that perhaps... or supporting things that perhaps don't have such clear-cut ROI?
Jim O'Shaughnessy:
I think that's a great question and we can go back to my example of a sculptor looking at using new materials that haven't been used in the past to create sculptures. Spending their time either developing the materials, studying the material, working with a variety of materials, seeing in an Edisonian way, finding 100 materials that don't make good sculptures. I am very open to what I would call a social return on investment. And look, we're going to get a mix of entrepreneurial fellows who want to do the heavy lifting over the year of their fellowship to create something that they view as a really viable and useful new company or product or service, that then obviously we would be their first choice hopefully, unless we really make them not like us.
And by the way, the terms of the fellowship is they don't have to use O'Shaughnessy Infinite Adventures. They can go to another VC is they prefer. We'd have our feelings hurt, but... As far as ROI, I think if the work that you are doing was going to provide either a body of knowledge or a new way of looking at your subject, or something that advances it where others can make use of your learning, that to me as a really high social return on investing. And look, I'm incredibly lucky in that I'm not doing this to add another zero to my net worth. It's like money has never motivated me really. And that's a weird thing to say because I've done really well, and asset management and finance bro and all that kind of shit.
If we see a fellowship application where we can clearly see that if the fellow's plan and study leads to a breakthrough or a clear addition to a body of knowledge that can be picked up and used by others to continue moving... It's like open source with AI. Why is it better? Because there's so many more people working on it and it's open, and they can use all of those tools. And suddenly, evolution and revolution become kind of the same. Things get turned around so much faster because you have cognitive diversity and you don't have same, same, same.
I would say that if you have a clear and compelling idea for what you want to do with the year of a fellow, and we could see that outcome might add significantly to society as a whole, that's great. We would view that as a win because that would be increasing our social ROI, and I'm all in favor of that. By the way, there's a fund, called the Acumen Fund, and they give money away like a VC makes investments. They get a ton of donations. Their whole pitch is you're going to get a fantastic social RIO.
Tony:
That's brilliant. And my last sort of micro question on that is you've obviously had a ton of applications and I feel like I'm already way behind the ball. Is it too late to apply or where would on this curve of-
Jim O'Shaughnessy:
Yeah, thank you for asking that question. It is not too late to apply. There is not a deadline for us to say, "Here are the 12 fellows." We are taking our time. We want to give everyone the respect they deserve, and that means that their applications are being read, they're being taken seriously, they are being discussed. In certain circumstances we have follow-up questions. And we're not going to pick a date and say, "Here are all 12 fellows." So the applications can continue on a rolling basis. I think it would be a mistake on our part if we had some sort of artificial deadline that missed somebody who had the greatest idea ever and they thought, "Oh well, there's already 500 applications, there's no way that they'll get to mine." We will get to yours. And so we take this very, very-
Tony:
Well, thank you so much. That's exactly the question in mind. I thought, "Oh, the boat has already sailed."
Jim O'Shaughnessy:
It has not.
Tony:
I missed the boat.
Jim O'Shaughnessy:
It has not sailed. And just also keep in mind this is going to be an annual thing. So if you don't make it into the 2023 class of fellows, you have an opportunity to hone your application, your ideas, everything else, and reapply for the 2024 class.
Matt Ryan:
Thank you so much. So next up we are going to have... Let's go with Mike. I know that he has a question regarding AI, so that's slots in nicely to Jim's favorite thing in the world. So Mike, whenever you are ready ahead.
Mike:
Alrighty. Hi, Jim. I'm a longtime admirer of your platform and business building. So thank you for sharing your time today. A bit of observation and context and then a question for you, putting our investor hat on. I'm a startup founder focused on search. Everyone in the world is focused on using AI and machines to do search. We believe that humans have special gifts and are re-centering the humans in search. And as you pointed out, there's like this Cambrian explosion right now and excitement around the AI space. It's available at a good price, sufficiently powerful, and the ecosystem is forming.
So right now as we look at what's happening out there, there's two big classes of innovation. The first one is sort of apps and people experimenting and taking what's available and putting it together into something new. And then there's sort of this other set of companies, like ourselves, that are taking this technology and going much, much deeper in beginning to run R&D processes around what is possible now. For example, now that transformers are sufficiently powerful, what type of neuro-symbolic representations can we extract or inject into these language models and vice versa?
When I say this type of stuff, and for this type of startup that's working on the R&D piece, what do you typically like to hear about that R&D? What gets you more or less excited? What types of concerns do you have when startups are investing that time, energy, resource capital to do that R&D? Perspective here would be really appreciated.
Jim O'Shaughnessy:
Sure. Well, Mike, I guess I would say that some of the greatest companies in the world that started the second tech revolution were R&D. I think that there's most definitely a place for a well-positioned research and development company that presents itself quite honestly to investors with a timeline, "Here's where we expect to have actionable items from all of the research and design that we are doing the deep dive on," and milestones.
And so I mean if you came to me for example, what I would probably do is set up milestone financing, in that we'd say, "Okay, what do you think you need to do your research and development at the highest level?" And you would tell us what that was. You would tell us what the headcount would look like there, what the cognitive challenge would be, and the different types of minds that you would want working on that. Obviously I would assume that you would have to have access to some compute. And another thing that I love about Stability AI is we are very open to sharing our compute. I always forget where we are in the league table. I think currently we're the ninth largest supercomputer in the world, which is 10 times NASA. Another reason that I loved the Stability AI model was that sharing trait. They shared a lot of compute time with a lot of companies that are doing very, very well right now. I would think about it from your point of view, and your partner's point of view, is what kind of milestones could we set up, and go to funder and say, "Here's milestone number one. Here's what we need. Here's milestone number two, here's what we think we'll need then." The reason that works is that people are sometimes really afraid. Let's say just for sake of argument... I don't know you and you haven't pitched me in the past, so anyone listening, this is not a setup. I don't know what Mike has got cooking here. But if you went out with a well thought out, well-crafted plan, both explaining why the research and development is vital to the emergence of the end- case that you're hoping to, and you didn't say, "Yeah, we need 20 million or 25 million," you're going to get basically nos, would be my guess.
But if you come and you say, "Yeah, so here's the first part with the R&D." You're going to have to probably offer some benefits to that first seed investor, that the series A and series B investors don't get, for taking the risk and funding the original stuff. But I think in smaller increments, it becomes a much more doable thing. If what you're doing... It sounds exciting to me, by the way. I think that it's always been one of my things that it is the combination. It's the centaur version of half human, half AI that's going to win. Kasparov proved that in chess, and we're seeing it emerge in a variety of other industries. And admittedly I have an inside view. But it's really interesting how it's not just games. It's a lot of things.
Mike:
I was just going to say, well first of all, yes, absolutely. ZackS chess, man and machine, human to machine, 100%. And thank you for these tips. And this milestone financing is a great way to think about it. In terms of telling that story and then telling that plan, right now there are so many things coming out that products can look quite similar, but what matters quite a bit is technically under the hood what's different, whether you're using foundation models or an LLM, whether you're building your own transformers or so on and so forth, own proprietary data. Would you suggest that founders out there begin talking a bit more about the technical? Or is that sort of going down a rabbit hole that investors don't want to have conversations about and is sort of a bad use of time?
Jim O'Shaughnessy:
So that's a really interesting observation. One of my jokes is that my primary purpose as being chairman of the board of Stability is to translate genius techie into human. And so they would not have said the ninth largest supercomputer, they would've said, "We have 900 A100s in a cluster." A lot of people don't understand that, and I get your reticence. I would only remind you that Google was, I think, the 17th search engine. They made all of their investors salivate because they demonstrated they had a technically superior search engine, and that was the game. Game, set, match. The low information VCs didn't get it. They wouldn't even take a meeting with the Google guys, because they all said the same thing, "Oh, there's Excite, there's Ask Jeeves, there's Yahoo. The world does not need another search engine."
Yes, it did. It needed a technically far superior methodology to the ones that were then currently in use. And so that's what got Google in the door with the smart VCs and with the smart angels. I think that you can make that pitch. I would use the example, it's yours to use. That's how Google got where they got. That's how a lot of companies that went on to be monsters in infrastructure, in a variety of things in tech, got where they got. They were technically better and they could prove it. And so I think that you should, as you build your pitch and as you build how you... Just A/B test it, man. Honestly, this sounds flip and I don't mean it to be flip, try it on your non-techie friends and try it on your family. Try it on somebody who is a Luddite. I don't know if you know any. But if you do know one, give them a go. It's amazing when you approach people and ask for their advice. And by the way, that's a little Human OS trick there. Never ask someone for their opinion, ask them for their advice. It triggers a different part of the brain in our Human OS. When you ask for advice, people are far more open because they feel that you will actually listen to them and actually want to hear what you have to say. When you ask for someone's opinion, people start immediately making it all about their pet peeves and their way of looking at the world. They'll give you not nearly as much as they would give you if you ask for advice. And make them open-ended questions.
Talk to people outside of your circle. I'm sure you've got a lot of smart friends who are not techies, but you might have one that is a literature wiz, or a history buff, or whatever. A cross-pollinization of ideas I think is something that generates just such wonderful synergies. It allows synthesis of knowledge that is not allowed when you get into just your deep niche. And believe me, I understand, having lived a lifetime in the several months that I've been involved with Stability, how deep that rabbit hole goes. And so when you're trying to communicate why it's important that you're objective, why you think you can do the research to prove that case, why if you get there that you will have a technologically technically superior product, and then try it on non-techies, they will tell you what they don't get.
And if they don't, ask them. I do that all the time. When I'm trying to figure out how to explain maximizing objective functions, I finally got it down to, "You tell the computer what you want the most of." That is not a correct use of maximize the objective function, but it's one that non-techies understand. And so, for what it's worth, that is the way I would try and tackle the situation. Make it understandable, make it such that you can... And don't hide the technical stuff. Put it in an appendix, because there's always going to be somebody that you're pitching that's going to read that entire appendix. A lot of them are on the spaces right now, and I know a lot of people like that. Don't strip it out, just don't make it your first page.
Matt Ryan:
Thank you, Mike, very much. We appreciate it. All right, next up we are going to have Reasoned Speculations.
Reasoned Speculations:
Hey, Jim, thanks for doing this. A big fan, and your intro was suitably chaotic. I've had an eye on the Wall Street Journal to see if ‘Jim Takes Over Another Small Democratic Nation’ shows up as a headline.
Jim O'Shaughnessy:
It's going to be a really small one.
Reasoned Speculations:
Go for Chile or something. I don't know.
Matt Ryan:
They've been through enough, come on.
Jim O'Shaughnessy:
That's true. We're going to leave them alone.
Reasoned Speculations:
Fair enough. My question is on, I guess, applications. You asked for a vision, a strong body of proof of work in one of the questions. For even the examples of the applications cited earlier, I'm just uncertain what that looks like. My idea is on more of the writing. I guess I have my own national accomplishments in the field, but that's not proof of work, that's just credentialing. I don't have a website to share, or examples of that or something. I'd just love some insight, thank you.
Jim O'Shaughnessy:
Sure. That's a good question. So then that becomes incumbent upon you to write a very persuasive argument for why you want to do what you want to do, with an explanation of that fact that, "I don't have a website, but here are my credentials." I'm not anti-credentials. Everybody falls into this deterministic, zero or 100 and, "Credentials are awful. They're the end." That's not true. I'm not going to a doctor who doesn't have good credentials. I don't want a submarine commander who doesn't know about nuclear submarines running the show. Credentials have a place.
And so if you don't have a website, and you don't have the work that you want to do, should you receive the fellowship, then write an excellent narrative defining what you want to do, explaining why you want to do it and why the use case is important. And like I said earlier to an earlier question, we are going to read every one of these and we are going to give them the due respect that we believe that they owed. People are taking a lot of their time and effort, and submitting applications which, like I said at the top, are incredible.
And just because you don't have a website or some other external proof of work, prove your work in your application would be my advice.
Matt Ryan:
Yes, sir. All right, our next person up is going to be, Dim. Dim, go ahead.
Dimitri:
Yeah, greetings from Finland. Jim, I wanted to ask you about apart from the fellowship, you also noted that you have Kyla Scanlon now as an entrepreneur in residence. I wanted to ask more about how that works, that part of your venture operations.
Jim O'Shaughnessy:
Sure, sure. That part of the venture is under Infinite Adventures. And unlike the fellowship, the entrepreneurs in residence has already developed a bit of an idea for what they would like to create as a company. In Kyla's case, it's this idea of her company that's called Bread. I had had earlier discussions with Kyla about Bread, where she was quite honest with telling me that she wasn't certain that she was ready to be a founder, right?
Dimitri:
Mm-hmm.
Jim O'Shaughnessy:
Because having great ideas is one thing, executing great ideas... Having started five companies, four successfully, it's not easy. You've got to learn how to deal with people. You've got to learn how to deal with a variety of those things. An entrepreneurs in residence is different from a fellow in that we will have our team work to assist her in her developing of her own ideas. Just tell me a little bit, if you would, about what you're thinking of?
Dimitri:
Well, no, actually it's related a bit to Kyla's idea in the beginning that she's trying to simplify really hard concepts into something much more simple. What happened to me that I actually got long COVID and I lost ability to focus for a while. Now I'm recovered almost completely, but that really scared me because I couldn't focus on a long time. I realized that a lot of people nowadays are in the same boat. Not in terms of COVID, but in terms of ability to do delayed gratification and everything has to be really fast consumption.
My question to you, my follow-up question, related to this was what do you think was going to happen in future when AI will be powerful enough to be used simply as a human machine? Do you think complex books like yours, like ‘What Works On Wall Street’, will be written by people in the future, or will be less of that? It will have to be more simplified?
Jim O'Shaughnessy:
Listen, I think that ultimately AI is going to be the most useful tool that humans have ever created.
Dimitri:
Fully agree.
Jim O'Shaughnessy:
But I think that it is a tool. And I think that humans will be freed in thinking about and approaching complex problems in a much different way than we had to in the past. And as the tools evolve, yes, AI, it will be super cool to be able to have the AI scan an array of 500 million data points. But we're going to be directing that. We are going to be inferring and learning from it and developing other questions and use cases. I don't think that generalized intelligent AI is something that we're going to see in my lifetime, certainly. I don't even know that it will be seen in my grandkids' lifetime because David Deutsch, who is vastly smarter than me, wrote an entire chapter about it in his book, ‘The Beginning of Infinity’.
To be able to program something, you have to understand the nature of it. And one of the key parts of consciousness is qualia. And qualia is that notion like if you have that first sip of red wine, the feelings that it invokes in you, or the memories, or the mood, if you will. We don't understand qualia in humans right now. And in fact, if you talk to some of the best neuroscientists, we don't really have a good working definition of human consciousness.
And so I think that for the foreseeable future, AI will remain what it has always been, and that is an incredibly powerful and useful tool to allow us to vastly innovate and speed up innovation in things like financial modeling. And also, and more importantly in my opinion, in medical research, in a variety of educational endeavors. I mentioned AI's use with the Synthesis School, but Stability itself has a mandate from the nation of Malawi to educate a third of their school children using our tablet and our Stable Diffusion foundational models. And the results, the very early results, are incredibly promising.
Listen, AI, in my opinion, is an omnidirectional technology. It will affect every industry. It will affect some industries different than others, but it will affect them all. I mean, for example, gaming, game engines, business services, presentations, publishing, advertising. The list is endless. Graphic design, copywriting, content creation, programming, product ideation. In finance, ways of modeling trying to understand financial markets better. In healthcare, protein folding DNA diffusion, medical imagine, predictive diagnostics, drug formulation.
I mean the list is literally endless. I do think that these innovations are going to have humans at kind of the center of it all. Not in all cases, like the use case that I mentioned earlier where we simply program an AI to generate a ton of useful null hypotheses. That, we don't really need. We can start the AI learning how to do that. But then we would hope that it could continue to learn, continue to generate better and better null hypotheses and just electronically publish them to a database that's accessible to all scientists. I see that as... I think that doesn't have to be just in medicine, it could be in creativity, it could be in creativity. I definitely see the day where I'll be able to create a book that is just about my new grandson, Charlie, or my new grandson, Jack, and I'll be able to work on it with them and it will be a book that is uniquely customizable to them. And by the way, you see there's a theme here, right? I was passionate and obsessed with customization. That's what led to NetFolio in the early days and then Canvas in the later days. I believe that we are entering a world where medicine just for Dimitri is a reality and that it is your genetic structure which we can now decode thanks to the pioneering work that took decades. We can now hyper-speed up a lot of this research. And the marvels coming are... I hesitate to even reach for examples because they sound so kind of crazy, but they're not.
Dimitri:
I've been really studying the AI, so I'm completely with you on this, it's true. We don't even understand what's going to happen.
Jim O'Shaughnessy:
And I think that's good though because one of the hiccups in human OS is this idea that, and we all have it by the way. I'm not excusing myself here, I'm just as running human OS just like everyone else. And you have to train yourself not to think this way, but we have this default assumption that we know everything there is to know, and we don't know one 10th of 1% of what there is to know. And so that requires a humility and a open-mindedness that does not come natural to us as a species. When you look evolutionary white, the hunter gatherer and the pecking order and all of that kind of stuff that's baked into our DNA and that's not going to be easy to change, but clever people who are willing to be open-minded about it and pursue these projects.
I think we are literally at the cusp of an exponential age. And the burst of creativity, innovation and everything else that regular people can both produce and take part in is an unprecedented period in history. And I just feel that I have been ridiculously blessed to be alive at this time in human history.
Dimitri:
I agree with you. So myself, I really got fascinated by your concept of a grey swan ever since we spoke about it, and it's constantly been in the back of my mind. And I listened to one of the way, I think it was a mud who told you that he'd eat half a billion to do it correctly. But I feel like, well it's now, but in future it'll be different. And you don't need to make the Hulk, that's too broad. You can also narrow it down.
Jim O'Shaughnessy:
Of course. And for those who don't know the gray swan, it was basically by the very definition of a black swan, which is it cannot be predicted. That does not exclude, can it be confirmed? And if you could confirm a black swan, like say oil going negative a few years ago, if you could confirm that 24 hours ahead of everyone else trading the oil market, you could make an absolute fortune. And so look, there's going to be a lot of that and there's going to be a lot of innovation and it's going to certainly reorganize financial markets, but medical markets, artistic markets, every market. And I think for the most part, that's a great thing because we're going to get better stories. We're going to get better models, we're going to be able to create more abundance. And I believe that we are moving into a positive sum world, and it's about time. We've lived, almost all of human history, in a negative or zero sum world, and we're no longer having to live in that world. And to get people to tune into that is another mission of OSB.
We're going to be very open about both our successes, but more importantly about our failures because we will fail and hopefully all the mistakes we make will be original mistakes and not ones that already happened. But I think that-
Dimitri:
I really admire your mission.
Jim O'Shaughnessy:
Well, thank you.
Dimitri:
And when you mentioned one of the episode that the bad guys already have the systems and you are the good guys to try. That really clicked me because it's so true. There is very good books about the advances of AI and the superpower of AI. AI Superpower, thing was called, one of the books. And it's quite scary.
Jim O'Shaughnessy:
And it is of course. But another reason why I'm so in support of open verse closed, is closed is going to lose. Why is closed going to lose? But-
Dimitri:
It's by design.
Jim O'Shaughnessy:
... going to lose because... Yes, by design open allows cognitive diversity and allows people of, we don't care what your background is, we don't care what religion you are or sex or what your politics are, what we care about is are you an innovator? Are you going to add something to this open system that makes it smarter? And there's that great quote, which I can't remember who said it, but it's very difficult to ask a person to come up with something that they can't conceive of. And different people can conceive of different things. And that's why open wins. It wins because-
Dimitri:
I have mine.
Jim O'Shaughnessy:
... markets work so well because a lot of the products that end up succeeding make no rational sense. And yet the person who puts them out there understands on a more intuitive level why humans might love this particular product. And so I'm obviously talking my book here, but I'm very bullish on all of it.
Matt Ryan:
Absolutely. Dimitri, thank you very, very much. Alistair, you are up next. Also, Jim, you and I will have to link up offline because I think Black Swan would be a great name for Martha Stewart documentary. So with that said, Alistair, why don't-
Alistair:
Yeah. Hey Jim. Hey everyone. Super quick one here. Just wondering, is there any kind of informal segmentation or ideal outcome you want to see in terms of how these fellowships land? We want a few in arts and culture, some on tech, some in CPG, is there anything like that or is it just totally wherever things happen to land? Thanks.
Jim O'Shaughnessy:
There is a little bit of that, but very little. I want them to land on the ones who have the best use case. I want them to be based on merit. But I don't want to give 12 scholar or fellowships to AI students. I want diversity in our fellowship because I also believe passionately in cross-fertilization across disciplines. And I don't want to award everything in even just the same field. I want artists, I want entrepreneurs, I want scientists. I want a nice representation across disciplines. And I think that that's important because there is genius everywhere, and I don't have any particular biases against a category and, or a subject other than I will not let all 12 fellowships go to a single domain. Other than that, we're trying to make this as merit based as we possibly can.
Thanks. Matt?
Matt Ryan:
Fantastic. Let's see here. Common, you are up.
Common:
My question is a tweak on the question I find you ask at the end of your podcast. And I wanted to ask, if you could incept someone who is the age of your grandkids now with the idea that they have to study one field, one discipline that you think right now is completely wide open for somebody to create a billion, a trillion dollar company and there's even less people than a hundred in the world who really understand it. What would that discipline and area be? Thank you.
Jim O'Shaughnessy:
Oh wow. See, that's so unfair because you try to nail me to just one industry. Well, I think that studying professions that will profoundly change with the advent of AI and other tools. For example, I think studying energy on the fission fusion divide and potentially other use cases in energy that we haven't even conceived of that potentially our experiments in AI will uncover might be a pretty cool thing to know a lot about. But who was it? One of the ancient poets, the proper study of mankind is man. I apologize for the gendering there, but that was written a thousand years ago I think.
So the challenge with that question, if I'm giving you an honest answer, my honest answer to that question is, that industry doesn't exist right now. And it's a bit like right before cameras were developed, training your son or daughter to be the best portrait artist in the world. And what happens when you ask a question like that is you immediately try to think of what everyone's talking about right now or what is a hot topic. You often, it's like, again, David Deutsch has a great line in his book, which is, "Hey, what were physicists saying about the internet and nuclear power in 1900?" Well, the answer is nothing because they didn't exist and obviously both went on to be massively important to human development. So I guess my answer is going to be unfair. I'm going to incept them to pay attention to things are brand new. Nobody understands the implications of that particular industry or innovation and then learn to master that particular thing.
But again, that bothers me too. I want them to do what they love, what they're passionate about. And me telling them to study energy or telling them to study meteor mining or how to create a livable environment on Mars, I think are all really cool things. But what if they don't want to do that? I couldn't incept my grandchildren and point them in a direction that wasn't their choice. They have agency and I am not capable of exerting my will over their agency. I want them to make these choices, not me. So final answer, Jim? Final answer is I wouldn't incept them. I want them to find their own path and passion.
Matt Ryan:
Fantastic Common. Thank you so much. Ahmed, you're up next.
Ahmed:
Hi Jim. I seen your podcast episode about the application for the fellowship and I wanted to ask a question around, I think it's the third or fourth question, where you asked about what you wanted to happen after the end of the fellowship. So I was wondering what is sort of the ideal outcome for the fellowship in your eyes of OS Ventures eye?
Jim O'Shaughnessy:
So as I said earlier, I think that the ideal outcome for me is sort of several things, right? Number one, I want to, as best as I can in my small little way, demonstrate to society, to people that there is so much brilliance in this world. And when all we're getting from legacy media is shouting and idiocy and our team is better than your team nonsense over 24 hours, to me that's torture. I used to say that if I wake up and I'm filling out a form and the walls are covered with 24 hour news, I know I am in hell. And so I want to demonstrate that that is patently false, that collectively, we have just such an amazing group of incredibly switched on, incredibly smart, incredibly good intentioned people in this world. And the way I can do that is to give them the funding that gives them the voice that leads to people knowing more about them. And this isn't just the fellowships, this is infinite films too. This is who we're going to fund through infinite adventures. We want win-win.
I don't know what you think about payday lenders. I think they're evil. I think they're predatory. I think they take advantage of people. And if you came to me and told me that I could get a 10X by investing in this particular payday lender, I would say no because they're predatory and that's not good for society. There's a lot of ways to do really, really well financially and add to as opposed to subtract from society and people. And so that's one thing. Another thing that I want to do is I want to demonstrate that it doesn't matter where you are anymore, if my friend Dan McMurtrie has a very successful venture firm called Anchor List Bangladesh, and I was just talking to him the other night and there's really exciting things happening there. And 10 years ago, I don't even know that I could have conceived that could be a workable company.
It doesn't matter where you are. If you have a great idea and you apply and we find you and fund you, you get the opportunity over those next 12 months to follow your passion and bring that idea into reality. And by the way, that's not easy. Everyone dreams but if you don't take action it's a daydream, it's a fantasy. You have to enter the arena. And the only way into the arena is to be brave. Being creative takes courage, man, because everyone's going to either be envious of you or poo poo you or say it'll never work. And there's an endless negative drumbeat that you have to overcome. And so it takes courage to go into the arena, but you got to go into the arena if you want your ideas to become real. You can dream and dream and dream.
And if you're sitting on the couch and you're not fucking doing anything, it's just that, it's a dream. It will never become a reality. And so by doing this, I hope to be able to demonstrate through these fellows, look at what they did, look at all these cool things that they did, and they don't have to be just companies. I'm sure that there will be a lot of entrepreneurs, and I'm sure that we'll fund many of them and hopefully we'll do well with them. And I take the other position there too, everyone focuses on, I don't know, Bezos or Musk or pick your rich person. Did you ever think about the other side of how many people's lives they made materially better? How many kids went to schools they couldn't have gone to if they had not worked for Musk or for Bezos or one of those people?
I know it's very unfashionable right now to say good things about these innovators, but man, I'll tell you, do you know why the government isn't suing Amazon for antitrust? Because the American people would be up in arms. They love Amazon and they love instant delivery. And some people hate it, sure, but the majority love it. So the next point is creating opportunities, creating jobs, creating interesting lives for people, for yourself. That's all on the plus side of the ledger as far as I'm concerned.
And then finally, we want people to know that it doesn't matter so much. We are moving from a world of tight networks to a world of loose networks. Networks are very, very important. Your network is your filter. I'm stealing that from Don Tapscott, but he's right. Your network is your filter. And if your network is just whoever you went to hub with or Yale or Stanford or Notre Dame or whatever, man you are in a poverty network situation, dude, because there are so many brilliant people in India, in Bangladesh, in Africa, in Estonia, in the United States, in Canada, they're all over the world.
And talent is, I think probably evenly normally distributed, access to funds hadn't been. And I think that the more people see this, the more other people... I think that this first group of fellows, I am going to commend them because they're the vanguard here. They're the tip of the spear. And as they succeed and bring really cool things into the world, other people will be far more likely to be willing to apply. But the final thing, as I said earlier, it doesn't have to be financial ROI, right? I think that that social ROI is equally delightful. And you never know going in. First off, you think, oh, well, it's just social ROI and that's great, and blah, blah, blah, and then all of a sudden there is a business idea out of it that no one thought of before demonstrating it on the social ROI side.
So that's what we're really trying to do here, and we're just trying to get and bring new ideas, new innovations, new discoveries into the world. And because it's going to, I hope, put a tiny little dent in the proposition that, oh, everything sucks. The world is going to hell. I see tweets where people honestly believe if you're having kids now, your grandchildren probably aren't going to live. And it just makes me sad. It doesn't make me angry as much as sad because you're building yourself a mental prison that is so limiting your aperture of you that you find what you look for, man. And if you see the world as a horrible and nasty place, guess what? The world's going to comply and it's going to be a nasty, horrible place for you. And it's going to confirm all your biases. If you see the world as a wonderful experience that where you have won the cosmic lottery, just breathing air, do you know what the odds against all of our being alive are? They're astronomical.
And so if we can just move just a little, the dial away from the pessimism, away from the tribalism and more toward, listen, what are we? Something like 97% identical genetically, and we're fighting over 3%. It's absurd. And the negative drumbeat, people do it for control, they do it for status, they do it for reasons that I am not in favor of for the most part. And even if we just move the dial just as scooch, I'll view it as a massive win.
Matt Ryan:
Thank you. Thank you so much, Ahmed, we appreciate it. Johnny Davis, you are up.
Johnny:
Hey, Jim, first, congratulations again. I'm really excited to see what happens here. Your last response kind of teed up the question I've been waiting to ask. Startup incubators like Y Combinator, you get access to this network, but it's kind of really collaborative. You have these four verticals that are usually pretty far apart, and you have your existing network that's pretty extensive. Are there any plans to facilitate a class between these verticals or a more intimate setting for them, like a private OSM community or OSV community?
Jim O'Shaughnessy:
Great. And this is Johnny, right?
Johnny:
Yeah.
Jim O'Shaughnessy:
Yeah. Okay, good. Sometimes when they do it by your handle, and that is, okay. So Johnny and I talk a lot, we know each other pretty well. Just full disclosure. Yes. The answer is absolutely. We're first going to build it in the ether, so to speak, but we will have discords, we will have all channels where we can interact with these various disciplines.
Somebody put up as a joke that I should buy some track of land somewhere, and then people started putting up, "this university is being decommissioned, you could probably pick it up for nothing." But honestly, that kind of appeals to me. It might be cool at some point, I'm not going to do it now. Something has to really hit big for me to be able to do that. But how cool would that be if you could just put together a community of creatives? But to me, it's got to be a varying disciplines because it plays into my cross-pollinization thesis, and that is... I actually got more insights on how to be a good investor from the Dao De Jing than many of the how to invest type books than I read. And broad reading across disciplines, it's really helpful because it allows you to synthesize ideas that are absent in one particular domain, but they dominate another.
So the answer to your question, Johnny, is yes. One of the things we're going to invite everyone who applies for a fellowship, whether they receive it or not, they will get an invitation to a community. And that's how we're going to start. But as you know me, we'll iterate, we'll change, we'll lean into what's working, we'll lean away from what isn't. But I'm a huge believer in these types of discussions because so many things get discovered when you're on a walk or you're talking to somebody who has no knowledge of your particular domain, and yet they say something and all of a sudden you're like, "oh my God. That's the answer." And I've seen it so many times in people I've worked with that it's not an n equals one in my case. And so it's something I really believe in.
And so the short answer is, you bet we are. And any ideas for how to make those places better, we'd love to hear them.
Matt Ryan:
Christian, go ahead whenever you're ready.
Christian:
All right. How are you, Jim? Just a question on, I'm living in Australia, Melbourne, I just applied. I actually had a tear of inspiration when I saw Dave Perell three or four weeks back share the fact that he wanted to do a fellowship. For me, I'm not really a startup founder. I don't want to raise capital for a tech company. It's probably the antithesis of where I want to go in my founder market fit. But I do see myself as a solo creator and one of my biggest frustrations is being in Australia, although we're one of the most livable cities in the world, Melbourne, the talent pool here for like-minded people is relatively low.
I find myself hanging out with a lot of high school mates, which is great, but at the same time the stimulation mentally at times I feel like I'm missing out by not being in America. So if I was to say win the application, is there an opportunity to move in vicinity to yourself, whoever else is in the community to kind of just build some, yeah, I guess closer proximity in that one year period? I'd be very interested in going to America or anywhere, just to kind of feel out a different culture because America is where a lot of it happens. So although I'm very grateful with Australia, just making that distinction. So that's my question.
Jim O'Shaughnessy:
Sure, I get it. I love Australia, I love Australians. I also love Kiwis and New Zealand. I've been to both many times. They're both wonderful places and countries filled with wonderful people. So yeah, of course. You can do whatever you want. If you can get a visa to come over here, yeah. If you need a plane fare, we could make that part of your fellowship. No doubt millions of people, millions, that's saying a bit much, but there would be no doubt lots of resources for where you might want to live. One of the things we are going to have, it's not for long- term residents, but we are going to have a clubhouse, kind of a guest house, an OSV guest house in the Greenwich, Connecticut area which is about 45 minutes from Manhattan. And it's designed more for shorter term, but we've heard some people that are already part of OSV, they're like, "I think I'm just going to come in the summer and chill for a month," and that's fine.
So that would be an opportunity should you get over here to America and you were a fellow, you could come hang out here and probably run into a lot of other fellows and other OSVers that we are working with. And by the way, I'm a huge fan of travel and getting to as many different cultures and countries as you possibly can. I think if you travel widely and talk to people of different points of view widely, it is almost impossible to retain a deep prejudice against anything because you see firsthand that at the core most people are pretty decent. And all of the things that the negative drumbeat trying to highlight differences, it's easier because we have so few differences as opposed to similarities, but travel really opens your aperture. So win or don't win, I would say if you can come over to America or it doesn't have to be America, be anywhere where things are happening, and in the meantime you have the ability to easily change your digital zip code and country code. And I'm sure you're already doing that.
Christian:
Yeah, definitely. Great answer. I've been searching on Twitter and I've probably DM'd maybe say 100 angels in the past two weeks, really just hitting and hoping that they would want to do something similar to you if I was to not be successful in my application. Why do you think that none of the VC funds or angel investors have actually thought about doing something similar to this? What is the gap psychologically in the marketplace that says, "I don't want to give an equity free grant to someone that's not your standard startup founder?" Curious question.
Jim O'Shaughnessy:
Yeah. So I think it comes back to a lot of people have to deal with partners or clients or LPs or what have you, right? And sometimes they want to try to minimize risk, and I'm not that kind of person. I am long ball, I love ball, and I am totally prepared for the idea that some things are going to be total clusterfucks, and I think that they're afraid of those. They're afraid of bad PR, they're afraid of their investors or their LPs thinking less of them or thinking they're stupid or calling them out or the forever, "I told you so," which of course is all retrospectively generated and driven by in hindsight bias.
And I mean, I'm not coming down on those people at all. Listen, I probably ... if I had to answer to a big group of LPs, I probably wouldn't be as aggressive in the things I'm doing. And so one of the great things that we can do with OSV is we have no agent principal dichotomy here. If I want to do, it gets done. And that doesn't mean that I'm going to be autocratic. I take the counsel of all of my team members very seriously. And there's several things that I ended up not doing because my team members were like, "Yeah, that's probably not a great idea." But because I don't have that agent principal problem, I have a lot more degrees of freedom than traditional VCs, traditional funders, and hey, I'm going to press the advantage.
Christian:
So do you think a marker edge for guys like Dave Perell or Jack Butcher are the solo creators that have what seems like infinite distribution, if they could almost attract greater deal flow and opportunities to JV with their community members or other creators in their community that a lot of these funds would just not touch? I see that could be a huge opportunity for some of these solo players.
Jim O'Shaughnessy:
Well, all I can tell you is there is one fund and it's called Infinite Adventures that will take a look.
Christian:
Yeah.
Jim O'Shaughnessy:
Because I'm open to all this stuff. Look, one of the things we're getting pitched on is ... I got to be careful. I don't want to identify it too closely, but it fits very, very much squarely in what you're talking about. And we think that there is a commercial possibility there, even though like a traditional venture fund, I doubt, I don't want to speak for them, but I would doubt would take a serious look at it. I think that there's a lot of really cool stuff coming out of your space, David Perell's space. As you know I did these scholarship for Write of Passage and we'll probably do another one.
Again, the quality of the interns that we got from David is off the charts. So I would not ... don't psych yourself out. Don't say, "Oh, VCs will never look at this, and therefore I'm not going to pitch them." Keep pitching, send this. I'll give you an action item right now. If you have an idea that you think could be funded, send it, your idea to pitch@osv. llc. Not guaranteeing anything, but we will definitely take a look.
Christian:
Awesome. Thank you so much, Jim.
Jim O'Shaughnessy:
You're welcome. Matthew.
Matt Ryan:
Thank you, Christian. We appreciate it. Mr. Hulk, you are up.
Investment Hulk:
Yeah. So I figure I'd just come in and stir the pot with some questions on the AI stuff, so.
Jim O'Shaughnessy:
Sure.
Investment Hulk:
I'm just wondering, thinking, and this kind of goes back, you hit on some of this stuff with the open source stuff, which I fucking love by the way. But also, and also fine-tuning training sets, you'll see me tweet about the GPT four chain model, which is GPT-J believe, then fine-tune on just the worst things humans have come up with and ironically it scores higher in truthiness than GPT-3 which says a lot about a whole bunch of different things. But yeah, just go on that, I know just because of stuff with stable diffusion or whatnot, I'm just curious with stuff with IP protection and that sort of thing and censorship, not safe for work content, how that stuff handled, the latent space and there's a lot of really interesting, dangerous, potentially beneficial things going on there, so I'm just curious your thoughts on some of the third rail questions.
Jim O'Shaughnessy:
Yeah, so thanks for it because I like to talk about it because I think people need to talk about it more, not less. Listen, there are all sorts of ... Look, can you imagine if they decided they weren't going to sell knives for people to use as cutlery because they can also be used to kill people? You can kill some human being with almost any implement that's sitting on my desk right now. And it's not the thing itself that is the problem, it's the human's use of it. And I think we have to accept the fact as an earlier questioner pointed out, the bad guys already have all this stuff, and I believe that this is a race of sorts and I think the good guys are going to win because I think open source is going to win. And open source is going to win because we can use 50,000 coders who are currently working on making stable diffusion better, more stable, et cetera, et cetera.
And so when you look at the ingredients of foundation models which is what you need to generate GPT chat, which is a large language model, stable diffusion, which I think is a superior model because it does both text and images, but multi models I think are kind of the future here. So you've got a self-reinforcing moat for the kids in the garage, right? GPU access, the price is coming down, but it's still enormously expensive and enormously scarce, right? So I will not go into the dweebie technical jargon of what it takes to have how many A100s in a cluster and the location of the clusters and proximity and all that, but it's not trivial stuff. Okay?
So that is a large barrier to entry. The second one, talent. Talent is really scarce. I am delighted to say that we, on a talent basis, our cup overfloweth and we're lucky because that was one of the things that really surprised me in a really good way. I bought into the, "Oh the tech overlords, all they care about is optimizing for money and everything." And I was really heartened to see people like David Hoff. I mean this man is a genius. He's the head of Google Brain in Tokyo. He's definitely in the top 50 minds of AI, and we didn't even recruit him.
He was incoming. "I want to come work for you. Who do I talk to?" And the intentions of these people are much better than I would've at first assumed. People really do have the best interests of what's best for society and the talent. So that's a small overlapping Venn diagram. Data, listen, a lot of data is noisy, unreliable and needs to be categorized correctly. You also have to have the ability to have models get smarter. To get a model smarter it needs new data, right? And there are a lot of proprietary data sets, think Disney, think Adobe, think the game manufacturers, where we will do foundational models as a service. Now those models will be private.
Open source doesn't mean we're going to be open about everything. It means that we're going to always release the most recent model open source so the developers have access to the most recent tools. But we do think that we will be doing ... The main part of our business is going to be these foundational models as a service that require forward engineers who require a lot of fine-tuning and require that the end user retain all the IP. We're interested in making the model smarter. We don't want their IP.
And so the access to data, you have to be very clear in your mission and in your behavior for these larger controllers, a very valuable IP to trust you, to maintain the integrity of their IP. And then finally the dollar costs involved here in training is ... they're really huge. And I would say honestly right now there are basically two platform AI companies that have all five of these things, or four rather, excuse me, and they're open AI and stability AI, and one is closed and one is open. One has a strong association with a big tech company. We don't currently.
Investment Hulk:
And getting stronger throughout the day and night.
Jim O'Shaughnessy:
Yeah. So look, there's room for both kinds. I'm in awe of some of the stuff OpenAI has done. I think they've done a great job and I'm not here to rag on them. Look, if they're adding to the collective knowledge base, which I think they are, I think should be applauded. I don't agree with their method. I don't agree that closed is superior. In fact, I think it's inferior for all the reasons that I listed earlier in the spaces. But hey, they got a bunch of money and good for them and they're doing good things. So I'm in favor of any of these companies that are going to add to the knowledge base.
To your specific question, I think we have to take all of the downsides very seriously and one that the public seems to be pretty worked up about and I think I understand why is this idea of deep fakes. And so Stability is going run a hackathon where the prize money will be in total $200,000. 100,000 USD is being put up by Stability, and because I believe it's important, 100,000 of USD is being put up by O'Shaughnessy Ventures. So the total bounty on a code that can identify deep fakes will be $200,000. We'll be getting that underway sometime this year. And if it meets our criteria and can identify X percent of deep fakes, we're going to give it away.
We're going to send it to all of the major news places. We're going to make the apps free on iOS and on Android. And we are going to try to disseminate it as broadly and as easily for people to get on their phone and then just wait until the first person who has nothing to do with AI has it on their cell phone and it comes up in deep fake. That's going to go viral. And yes, will the deep fakes get better? Of course they will. That's the nature of how AI learns. But we hope that we'll have a dog in the hunt, right? Because that's part of our mission. We also think that the discussions about what's ethical, I think that's a loaded word because ethical is different to different creeds, it's different to different types of philosophy. But I think that the best way to facilitate a conversation that actually gets somewhere is to ... I want an open wiki on it, on the ethics of AI that is moderated like Wikipedia is so you can't just have people in there flaming and being trolls. But again, in the spirit of why I believe open AI is going to win, I think an open wiki is also a good idea because I don't know, I'm sure that there's some 16-year-old out in his mom's basement right now who's 10 times smarter than I am about this stuff and has a really great idea for putting a constraint on a model that I would've never thought of, and if he had a wiki or she had a wiki that she could put that on, that's great.
So we are also putting together a council of people who have expertise in this field. We're going to try to be as merit-based as possible, but here's another example of credentials do matter, right? So we're going to get some of the top thinkers in this field on, hopefully they'll join the council. We just think sunlight is the best disinfectant, right, and so I always get queasy in my stomach when somebody tells me that they're doing something for my safety, for my benefit. I think that's bullshit most of the time. And what they're doing it for is they want to control me and they want me to behave the way they think I should behave. And I don't truck very well with that. I do truck with open conversations that are not limited, and I love the open Wiki idea.
I think we would generate a lot of great ideas. And so we are definitely thinking about this. We also listened to our users and we took a lot of heat for the release of Model 2. ... We're at 2.1 now, so I think it was either one, whatever. But we ran a not safe for work filter on it. And we did that because we heard from a lot of our users that we should do that. And I just believe in flexibility. So I believe that AI should be programmed in every language on the earth. And I believe that different cultures are going to have different taboos and different hot buttons that they should be able as a culture or a country to augment the model in a way that suits their purposes.
I may not agree with it, but who am I to tell a Hindu that he or she can't have their beliefs? I don't think I have that prerogative. And so by putting things at the local level, which is what we're trying to do, we want to set up JVs in every country so that we get the local flavor and plus we're not doing this just to be philanthropists. We're going to get more data and we're going to get data that other people aren't training on. And guess what? That's going to make our models smarter and the ones with the smartest models are going to win. So those are-
Investment Hulk:
You see a huge amount of that with the open source program languages kind of taking over the world in the past year, same principles that worked.
Jim O'Shaughnessy:
Totally. And it works across ... Ballmer called Linux communism. And of course he did because his paycheck made sure that he had to have that point of view. Listen, open source, just I think you're going to lose the debate if you try to make the case that open source, where you provide massive cognitive diversity, massively different skill sets, is going to lose to closed source where people think they can pick the winners ahead of time. The record on that is miserable and it is self-evident, not just in AI, in software in general. It's evident across many disciplines.
And by the way, I'm not saying you can't have closed source. I'm saying that if you want the models to scale quickly in the best way in which they can become smarter than all of the other models, then you are going to default to open source and people, "Well, you're not going to make any money on that." Well, Red Hat disagrees. I mean there are foundational models that will be a huge revenue provider. As long as we maintain our view that we will release the latest model with the weights to the general public so that other developers, other coders, other people who have ideas very different than ours can use it to build apps, to create all sorts of things I haven't even thought about, I'm sure a lot of our techs haven't even thought about, I'm going to be happy with that. That doesn't mean we're not going to do foundational models as a service. We are. And you've got to have a good revenue source to keep the company going.
Investment Hulk:
I mean, some of the large corporations won't even let cloud services on site. Basically you have to bring in ... I know of at least one company, I'll put it that way, where even getting GPUs on-prem is drama and that-
Jim O'Shaughnessy:
Oh, believe me, and I have it on excellent authority that a very well-known company in tech has many of the features that their client base would love in the cloud are not in their cloud. They're not even in their cloud. They're hard coded on the ground at sites that they don't talk about. And so-
Investment Hulk:
I think I could figure out this company.
Jim O'Shaughnessy:
Yeah, I think you probably could.
Investment Hulk:
Yeah.
Jim O'Shaughnessy:
But I think the idea though is when people realize that they're putting themselves at a severe competitive disadvantage to people like us, they're going to change their ways or they're going to die.
Hulk:
And also just in terms of network effects and moats and just survival in the modern software world, I mean open sources, I don't think there's really much an alternative for success to be honest. Especially when you start models that use models and when this thing really starts going, I mean, honestly, just even with chat GPT, I don't think there's really been a huge amount in terms of functionality versus some previous models, but it kind of hit certain usefulness thresholds where it just jumped up to a whole nother plateau. Even just some of my personal stuff with web development or certain holes I've had where I've had to use ... a need to do certain things where I wasn't able to, an organization I was at previously or whatever, it's like I've probably learned at a faster rate in computer technology the past month than I have at any point in my career. It really, really helps.
Jim O'Shaughnessy:
Very cool. Very cool. All right, Matthew, let's have one last one. What time do we got? This was supposed to be an hour and we're already at an hour. It's like my podcast. We just keep going on and on and on. So let's do a final question from the audience, Matt, and then say au revoir.
Matt Ryan:
So first person on deck is going to be Doggy Capital. You can go ahead and unmute yourself. And then Michael, you're going to be up next. I'm going to pop you in a speaker, put you on mute, and when Doggy Capital is done, you can go ahead and pop in and then we will go ahead and wrap for the evening.
Doggy Capital:
Good. Good. Thanks for staying on so long. I just wanted to ask you about real quick because I haven't heard you talk about it much, maybe you did earlier, is what are you trying to accomplish with Infinite Films? What are you looking for? And what's the workflow that you're going to be doing over at Infinite Films? What does that look like?
Jim O'Shaughnessy:
Very good question, and we actually didn't talk too much about Infinite Films, so I appreciate the question. Listen, I've always been a huge movie buff. I love movies. As a hobby for most of my life, I like writing fictional treatments, some thrillers, some whodunnits and that. And I've always loved the movie industry. I have a lot of friends who are directors. And I'm just fascinated by the process.
And I just got tired of everything being a copy of a copy of a copy of a copy. And I got tired of the fact that Hollywood felt that its new mission should be to beat us over the head instead of introducing any nuance or subtlety, here is what you are meant to think. And most of it wasn't inspirational. It was a bit schoolmarm-ish, in my opinion. A lot of tongue clucking and you're a bad person, et cetera, et cetera.
And I grew up on great inspirational, fun movies, funny comedies that were a lot more loose in terms of that. I think specifically though about the dearth of inspirational movies. I use ‘Rudy’ as an example, which is a movie about a kid who is desperate to go to Notre Dame and play for the Fighting Irish, even though he is five foot nothing and weighs a buck nothing.
And it's his perseverance, his determination, his work. That's the other thing, they make a big point of the fact of. He goes through deprivation. He sleeps in cots in the maintenance area, he runs and works jobs and finally gets the job just to suit up. That's it. They put him on the field and he got to be pictured with the team, but he wasn't a big player. But it was just this little guy got on the fighting Irish team. And like, my wife and I say, it's tingly mo moments in movies and it's ton of them. And it's a great inspirational movie for people who might feel like, "Ah, I'm not big enough, I'm not smart enough, I'm not this, I'm not that."
And so for our first movie, it's going to be a documentary about Dr. David Rhoiney, who I discovered when I did an experimental NFT. And the guy who bought it was a big believer. He is a very successful software developer. And he also happens to be a big believer in rising people above the radar as he puts it. People that he thinks people should know more about, he wants to get them some attention.
And when I listen, there's a podcast. You can see it at or listen to it at Infinite Loops. And man, I got to tell you, his story is 10X Rudy. This is one of the most inspirational men that I've ever had the privilege to meet. And I spent a weekend with him and his family and this guy is just amazing. What he's overcome in his life is just unbelievable.
And it's not that he's just a robotic surgeon. He's a savant at cryptology. He's a graduate of the Naval Academy where he played two sports and was there on scholarship. I mean just what a story? Our first movie is going to be about him and it's going to be a documentary. The workflow for the first one will be for people who have their aspiration is to make documentaries, write documentaries, work on documentaries.
My guess is that we'll be using mostly fresh out of film school talent. And we're designing it so that it's a win-win, meaning that anyone who gets hired onto the project will own a piece of the movie. I'm pretty good at selling things to people. We are going completely outside of normal distribution channels. We are only interested in streamers or platforms like YouTube. And we think that we'll be able to sell the films to streamers. I've already done a little homework there and it sounds like they have a pretty strong appetite for this kind of movie.
And then, after we sell it, everyone who worked on it gets a cheque. But more importantly, you're also going to get the credit. How cool would it be to be 25 years old, graduate from one of the better film schools and a year later, be listed as the director of this cool movie about this amazing man? The ability to amplify somebody's talent and career is really exciting for me. And so, the workflow will be different for different movies. If we do another documentary, we might get the same people or maybe that man or woman who was the director gets snatched by Hollywood. I hope they do.
But at the worst case scenario, they'll get a notch on the Screen Actor's Guild card. You need three appearances. Or let's say on the acting side, to get a SAG card, you need three appearances in films. I don't know what it is on the director side, but they have similar rules. Hollywood is still run a medieval guild association.
And if I can help younger people get those things, those check marked checked, that's a win for me. Even if the movie ends up being horrible, I'll take the responsibility for that. I just see again, there's just so much talent. And especially with younger people, the tools that are available today to make unbelievable movies are incredible, man. And I have a friend that's a director and he showed me a short of a movie. And he was like, "What do you think?" And I said, "Looks like a cool science fiction movie. And he goes, "Yeah, it was sent to me today. It was made entirely on an iPhone and edited on the iPhone." And so, I'm not a digital native, younger people are. And so, it's not to say we're going to exclude if there's somebody who wants to make a big career change and he's 55 or she's 55, we're not going to say no.
Those are the kind of movies we want to make. Those are the stories that we want to promote. And then ultimately, I suspect that maybe in three years from what I know by being Chair of Stability AI, this will probably be a mixed use film vertical where we do a lot of AI generated movies, but we're still going to need people to write the screenplays, to do the creative inspiration. And so, the work patterns will change, but they'll always be work.
And my best case scenario is you make some money, you get a lot of experience. And you get pushed out into the community where you might not have been able to get to without the association with Infinite Films. That's our game plan there.
Doggy Capital:
All right. Cool. Thanks, Jim. And just one comment on your tail end there. With the AI and visual work, I think the AI disrupting the visual effects industry and film and television is going to be huge. Who knows what that is. But you see the video game engines, like Unreal Engine 5 by Epic and Unity, they're not using AI, but they're trying to make it easier to produce these things. And they're getting very close.
And whoever's able to undercut what the current workflow is everybody's going to be coming to them. Because the quality of what can be produced and how much cheaper it can be produced, the possibilities are big. Hopefully, you can be the guy who helps facilitate that. But anyways, thank you for your answer. And I'll let you get to bed and answer the next guy.
Jim O'Shaughnessy:
I appreciate. And by the way, thank you. I agree with your comment by the way. All right, Matt.
Matt Ryan:
Fantastic. Hey, I'm on it. Okay, Jim. Give me a second. Thanks. Michael, you are good to go.
Michael:
I'm not sure why Investment Hulk doesn't scream while he speaks it. It's just out of character. Two, I've applied for a job as an intern with Ramp, but since he's now mixing water into his egg mixture, I've never gotten a response and I'm so disappointed. But that being said, thank you for allowing me the opportunity to speak.
And the question that I have is, I'm working on a project which is geared towards reuniting fathers with their children who've been separated via litigation. Now, there's someone in the audience who's near and dear to my heart. We live to serve. We've had a couple of conversations over the course of the year, starting end of last year. And I spent this last year reuniting with my son. And I put up a site, it's just a giant brainstorm and I'm going to take it live now.
Jim O'Shaughnessy:
Cool.
Michael:
I would like to ask your advice. When dealing with a legal system that's primarily funded by adversarial proceedings, if you have some sort of new development, which allows pro se individuals who can't afford attorneys and can't afford the blistering years of litigation. And you develop a tool and you push it out, how do you deal with the pushback from the legal community that says, "Hey, you're taking food out of our mouths. We went to school for many years and we built these practices," and so on and so forth?
Jim O'Shaughnessy:
Yeah. Well, first off, congratulations on the effort. I think reuniting parents of either gender who've been not allowed to see their children is both bad for the parent and bad for the child. Unless they're abusive or have some exogenous variable, which in this case, it doesn't sound to be the case.
Michael:
No.
Jim O'Shaughnessy:
I think that if we could work towards moving these types of disputes to arbitration as opposed to the traditional, let's duke it out in the courts, that would be a beginning. It's not an easy thing. But to be able to move to a simpler system of adjudicating these disputes that do not involve high price lawyers, do not involve all of the roadblocks that our legal system puts in the place of, say, a dad trying to reconnect with his son or daughter, that would be a positive.
One thing that we've been experimenting with is the idea of jury systems that get put together ad hoc. They were originally suggested for reviewing political initiatives because, as I said earlier, most people share ideas much more than they share disagreements. And when you put together a group of regular citizens, the urge to work together to come to an equitable solution actually has been proven in the various test cases where they use a jury to adjudicate a problem. That could be a potential thing. I don't know, maybe a nonprofit could sponsor such a mechanism. I think if it were me trying to design a story to get people to understand and back it, I obviously would come at it from an extremely emotional point of view. Kith and kin, I mean, that's the base code of our DNA and the bond between a parent and a son or a daughter is incredibly strong. And it's very, very destructive to artificially disallow it. I think that you could get, if you positioned it right, I think that... You say you have an app that allows this?
Michael:
No, not yet, but there's a piece of software.
Jim O'Shaughnessy:
You're working on it. Okay, I get it.
Michael:
There's a piece of software in development which is going to allow an individual on a pro se basis. If they receive a complaint or they're served with let's say divorce papers or custody papers, they're able to, instead of going out to hire a lawyer right away, they could scan that app in via an ML component. And the software will tell them, "Hey, this is what this is, and your options are A, B, and C. Which direction would you like to go in?"
Jim O'Shaughnessy:
Oh, okay.
Michael:
There goes the workflow.
Jim O'Shaughnessy:
Oh, okay. Okay. I misunderstood your question. Actually, I got pitched on something like that, and I'm sure it wasn't you.
Michael:
No.
Jim O'Shaughnessy:
Because, yeah, it was slightly different in the presentation. That's going to be, I think a huge app and I'm glad to hear you're working on it. The idea of people who can't afford a reasonable legal representation, it screams for ML apps that allow them to have explained to them in layman's terms what their real options are. Go ahead.
Michael:
Forgive me for interrupting. What I have up there is basically a giant brainstorm of all the services that let's say a mom and, or a dad would require in order to stabilize and facilitate the process of reunification thereafter. That's what exists now. Hopefully, within a couple of weeks we'll see something different.
The ideas will still remain, yet the technology needs to be pushed so that people can start testing it and seeing if it works for them. I'm not an attorney. I don't represent anybody. It's just, I've been through the mill. And after seeing all of the things that I needed access to, which didn't exist. And then I had to piece together a solution, I was like, "Okay, this would be great if we can get 10 dads and put them in a small community." Boom, same thing for moms.
And then provide them with the litany of, or the portfolio of services rather, which says, "Hey, here's one, two, and three. And then here are the remaining that you could choose from and see if you create a better life for yourself. It's all in your hands." That's what's there.
Jim O'Shaughnessy:
Yeah, I think it's a noble endeavor. I think that you're going to see a lot of apps that will aid you in this, because we've seen a lot of legal apps in development. They haven't been released yet, but they're really sharp. They're really amazing.
Michael:
What have you seen?
Jim O'Shaughnessy:
I can't name any company because we signed a couple of NDAs. But I have seen an app where I'm demonstrating it, even though I know you can't see me. I picked up my cell phone, okay. And I hit the app and I explained my situation in plain English. And I say, "I've been arrested for shoplifting. I contend that I wasn't shoplifting. But the officer who arrested me said this, this and this. And he said that they were charging me under section 543." I'm making all this up. "And I have no idea what to do." And the app immediately goes, "Here's where you have a case. They have...
Michael:
Shut the fuck up.
Jim O'Shaughnessy:
Yeah. Exactly, right. "And here is the section that seems to be misapplied. Here are the questions you should ask the arresting officer." It's really slick. And it comes with a recording function so that you can record the conversation that you have with your arresting officer.
And then it analyzes what was said too. It's really cool. And it's cheap. But that's just the one I happen to see recently, I've seen others that answer real estate questions. I've seen others that are just general legal questions like LegalZoom. I'm sure you're familiar with that.
Michael:
I'm familiar.
Jim O'Shaughnessy:
And so, I think you're going to probably see come online in the next 12 to maybe 15 months, some pretty slick apps that will work on your cell phone, that you could use in conjunction with your efforts. And I still would make the appeal to people emotional because it's a very emotional topic. And there are a lot of people in the boat that they don't understand this system. It's a Byzantine system. And that's yet another great use case for AI. It can break that stuff down and explain it in normal, regular language so that people can take action.
Michael:
All right. I appreciate it. Thank you so much for taking the time to answer the question.
Jim O'Shaughnessy:
My pleasure.
Michael:
And as Matt said, I have to make it spicy. I'm going to leave you with these two things, if you so choose to answer. How do you prompt? Number one, what's the best way you find prompting? And two, ala Rick and Morty, would Rick give a shit about any of this? Thank you so much, Jim. Good night.
Jim O'Shaughnessy:
Those are two really great questions. How do I prompt? I have a headstart because I'm 62 and I have a pop culture memory that younger people don't have. And so, I know movies like ‘Blazing Saddles’ and ‘Young Frankenstein’ and TV shows like ‘Mash’ and the ‘Six Million Dollar Man’ that a lot of young people have never heard of. And so, I prompt for those more exotic things that people often don't remember. Or if they remember them, they only remember them vaguely. And oddly enough, some of the GIFs and memes that come up from them are absolutely fantastic.
Rick and Morty, would Rick give a shit about any of this? Absolutely not. Rick would tell you that the universe creates humans to consume them. He would basically say, "The universe designs human as basically things to eat and dispose of." But I love Rick, nevertheless, because he is incredibly funny and also profound in many respects. ‘Rick and Morty’, I think could be taught as a philosophy class.
I once did a salon for inter-intellect, Anna Gat’s organization, which I think very highly of and have a small investment in. And the whole thing was Jed McKenna and ‘Rick and Morty’ explained the universe. And it's a little bit like Doug Adams, who wrote ‘Hitchhiker's Guide to the Galaxy’, which I love. And if you think about Adams more as a philosopher and less as a comedic sci-fi writer, it all begins to make so much more sense. I feel the same way with ‘Rick and Morty.’
Final thing I'll say about ‘Rick and Morty’, he calls it a story wheel, but it's really the hero's journey that Joseph Campbell made quite famous. He modified it a bit, but that's why it's such a compelling show. He manages to shrink down the hero's journey into, whatever, 24-minute episode. And it's just brilliantly written. But to answer the question directly, yeah, no, Rick would say, "Fuck you, I don't care."
Michael:
Thank you very much, Jim.
Jim O'Shaughnessy:
You're welcome. Matthew.
Matt Ryan:
Yes. And with that, I would like to briefly break character, unfortunately, to thank Jim for taking time out of his, what I can only imagine is a very busy schedule of jet setting and sitting on yachts. And I don't know, shooting endangered pigeons. I really don't know what he does. I just know that it's very esoteric, eccentric, and slightly amazing.
And thank you, honestly, to everybody who had questions for Jim tonight. In any kind of a scenario like this, the quality of the questions and the quality of the answers is really what makes the experience. And as many of you know, because I was DMing you while we were doing this, setting up the order and everything, this was fantastic.
The end product we got out of this because of the quality of the questions, and Jim's somehow amazing ability to string together coherent sentences, was just really fantastic. Jim, I'm sure you will probably want to do more of these in the future. We will keep everybody posted on that.
If anybody had a question that did not get answered, by all means, please, please, please, please reach out via DMs on the OSV account. I will personally make sure that we get an answer back to you from Jim. I've got enough pieces of work sitting out that he wants to see that I can hold those hostage until he answers it. I have got you guys covered there. But just thank you, everybody, for taking the time. This was... I don't know about anybody else, I had a fantastic time.
Jim O'Shaughnessy:
I would echo what Matt said. It's really the quality of the questions, and you had great questions. Thank you for giving us your attention and your time. Your attention is your most valuable resource. And we do truly appreciate that you gave it to us tonight. We will do more of these, and we thank you for giving us the gift of your attention.