[0:00:00] Thank you so much. Thanks everybody for having us. We're excited to be here. And I thought this would be an interesting conversation to talk about things that Sam and I don't necessarily talk about publicly as much. So to give us an opportunity to share thoughts about AI, its impact on humanity, its impact on education.
[0:00:29] What the education system needs to do in order to prepare humanity for this world. And so I'll be joined by Sam in a minute and we'll explore those topics together and just see how the world of education, the world of AI will be connected. I can hear you. There he is. Welcome, Sam. Thank you. Thanks for having me. Great.
[0:00:59] I thought maybe to start the framing and the conversation is to just acknowledge, and I think Steve mentioned this in the previous exercise, which is predicting the future is really a futile exercise. I'd much rather talk about the kind of future that we're hoping to build and talk about the impediments, the barriers, and the opportunities
[0:01:28] that are associated with it. So one note of caution is by looking at the kind of invented futures that have gone in the past and what has actually transpired. One example of it, of course, is the introduction of search and Google organizing the world's information. And I think at the time, we
[0:01:54] Who are in the sector, we're thinking about, oh my God, this is going to be such an amazing thing. We're going to have the machine organize all of the world's information. And that will mean that human beings will no longer have to do the drudgery and instead can focus on higher order thinking. That great editors will emerge and really leverage the power
[0:02:17] And I think it's relatively safe to say that editorial prowess around the world is not exactly accelerated. We've seen the world of top 10 lists and click bait and listicles. And so, as you think about, I don't know, reflecting on AI and
[0:02:44] It's immense opportunity to accelerate human potential. I don't know. Is there something that you're worried about in having an unintended consequence where instead of elevating humans to do better and better, that it creates a, well, it's better than average and that's good enough type of acceptance and culture? Yeah. Look, it's a great question.
[0:03:09] We are like any powerful tool. People will use it for good things and some bad things. And even though we're confident that the good things will outweigh the bad ones by orders of magnitude, that doesn't make the bad ones go away. And with any technological revolution of this magnitude, some important things are going to be lost, for lack of a better word. That doesn't mean we shouldn't do this whole thing. The benefits are going to be so tremendous.
[0:03:40] And I also find it very hard to predict the future. So talking about where the incredible benefits and the downsides will be, that's the only thing we can say for certain is that everything now will be wrong. But we are very aware that those are going to come and be part of it. And we'll work as hard as we can to minimize those, but not everybody will agree what the bad parts are even. And this is going to be a moment.
[0:04:03] That we're going to have to navigate together. The collective preferences of society is an easy sentence to say. Yeah, for sure. You know, and I don't know, it just seems like when we think about how do we instill values in society that reaffirm the benefits of working hard, the benefits of leveraging, as opposed to
[0:04:32] Really saying, well, hey, it's done so much easier than we can just ignore everything else. There's this great little cartoon in MySpace in the education world that shows a robot lecturing to a group of other robots and effectively saying, hey, great, now we can outsource everything. The professor will be replaced by AI, the students will be replaced by AI, and now nobody needs to do anything.
[0:05:01] Obviously, I don't think that's where any of us are actually heading. Certainly, I don't think where we want to be heading. But how do we actually prevent the forces that, or try and prevent the forces that are going to get us there? Well, I think there's a huge role for governments of the world to play. And I think there's also just a huge role for the kind of global discourse on this. One of the reasons we wanted to release ChatGPT is, I think this stuff is,
[0:05:35] Easy to talk about in a theoretical vacuum, but very hard to make progress on. And we were kind of saying all the same things before the launch of ChatGPT, and people didn't take them that seriously. And now people can feel it viscerally. When you can feel something, you can kind of develop intuitions for it. Also, I think by incrementally, gradually putting this technology out into the world, people have time to adapt and think about what they want and debate ideas.
[0:06:05] The very global conversation that has happened in the 10 months or whatever it's been since we've released ChatGPT, I think has been one of the most important things ever to happen for the progress toward this good future with AI in it. Yeah, I couldn't agree more. I think the kind of discourse that you have enabled in the world, I see it in higher education,
[0:06:33] The realization that things are going to change. And you have this battle between those that say, well, things are going to change. And so how do we avoid that change? As opposed to people who think about how do we embrace it? And I remember a conversation you and I had, my God, it was about a year and a half ago now, right? Where you were mentioning to me that at the time, I think it was, correct me if I'm wrong, that 70% of the code,
[0:07:03] OpenAI was actually being written by the AI, right? And so you had this dramatic acceleration of the productivity of your engineers. It's very amazing what's happening with ChatGPT for coders or Copilot or any of these tools. But I do want to point out that the code that is being written is the kind of, generally, it's the simpler kind of more road code.
[0:07:32] But this is wonderful because it's like, I think it would be wrong to say, hey, AI is doing 70% of the cognitive work of a coder now. But what is happening is AI is freeing up a dramatic amount of the time of a coder to work on the hard part where humans have to figure out what to do. Exactly. And to me, that's the most profound thing, right? Because, you know, we see it in coding in the...
[0:08:01] Leading edge. But think about then having not just a 10x engineer, which we all knew pre the world of AI, the mythical creature, which I've employed a few in my past, that could do the work of 10 engineers. But also think about the 10x lawyer, the 10x reporter, the 10x author. And what that kind of productivity can mean, especially
[0:08:30] For those that are going to have the training to have the metacognitive tools to use these things effectively. Because I think exactly as you're saying, you're able to do so much of the low value, but substantially time consuming work that people had to deal with. But it's grappling with the nuance and the black versus the gray. And it's those tools that
[0:08:59] I think obviously becoming more and more important in society, and frankly, that educational institutions have talked about for a long time, but really certainly up until the age of Minerva have not actually been doing. Yeah, I think we're all still underestimating that kind of impact and how dramatic that's going to be. You can see it the most in young people who, like, again, trash PT has just not been out for that long.
[0:09:28] But the people that have had the most sort of like plasticity in their workflows have, so like, you know, people who graduated college say last year and just joined the workforce without a set way of doing things. It's amazing how much they've adapted what they do to use this tool. Same thing with, you know, people learning right now. It's just remarkable. And I think that's a good preview for what's going to happen. Absolutely. I see it in our students. You know, the rate of entrepreneurship,
[0:09:58] In our graduates, we were already, I think, somewhere around two orders of magnitude, 24 times, I think, the rate of entrepreneurship among a Minerva graduate compared to a Stanford graduate, as an example. And now, I see them, just even students that took a different pathway post-graduation, saying, wait a second, now actually starting something is so much more straightforward.
[0:10:28] I have so much leverage, I need to raise less money. I remember back in the dark ages when I was starting work in the internet in the late 90s, we had to have a team of 20 engineers spending, I think, four or five months building a shopping cart. You don't even need AI, you just have libraries that you can use now and you can have one overnight.
[0:10:57] The ability to leverage that is so crucial. And because educational institutions are areas that really emphasize just rote knowledge and being able to pass tests, they haven't really equipped the traditional student with all of the things from their course of study. And so therefore, students are just figuring it out on their own.
[0:11:22] It's not that they've learned how to do this in class. It's that they've just adopted this incredible tool and have adopted it as their own. And what we're trying to do is give them these systems of thinking so that they can purposely leverage it. And in many ways, think about those types of
[0:11:43] Not a founder. I ran it for a while.
[0:12:08] But that's got to be, you must see the nature of startups also shifting because of the technology. Yeah, look, I have a group chat with some of my friends and we have a little betting pool going on about what your first single person company that becomes a $1 billion company will be started. As these tools get more powerful, what a single entrepreneur can do is just amazing. Yeah. So what, where's your over under? What's your bet?
[0:12:38] I don't remember the number I put down, but it's still a few years away until it starts, and then it'll take a while to grow. Yeah. Okay, but now let's think about that, right? Because I think it's...
[0:12:51] In general, I remember, again, one of our bets where you and I, I'm not going to repeat it because that's in the geopolitical sector, but you and I were the only pessimists about the use of a certain technology in warfare a while ago where everybody else thought that would be an absolute impossibility, no way. So I resonate with your instincts there. But given that, given this concept of literally the 100x,
[0:13:21] Productive human. How do you think that the whole education sector needs to adapt? I mean, if you had your druthers and you could wave the magic wand, what would you want to see educational institutions focus on to get the graduates ready? Because whether or not the tool is going to be ready two years from now, five years from now,
[0:13:51] Early stages of the career of every single person who is being educated today. I think the main thing I'd like to see right now is just a lot of experimentation. No one knows the right answer, which is fine. But I think systems, countries that have the highest rate of experimentation are most likely to figure out how this works. I think it is going to change the process of learning, no doubt.
[0:14:22] But we're still in this Cambrian explosion phase and people are building amazing applications with the technology. But we need to see where it's going to go. Yeah, I think it's such an interesting insight. I was last week in Mexico where we launched a brand new university, Universidad de la Libertad, which was actually privately funded. So it wasn't actually the...
[0:14:50] National initiative, but for the great credit of both the team and the government, the government authorized and accredited a brand new institution with a radically different approach, specifically in a post AI type of approach for it. I was there because of course we helped design the institution, but the ministers of education that I've talked to that are
[0:15:19] Trying to embrace fundamental change are crucial. At the same time, there are a number of university presidents. I'll give you two little anecdotes. I'm just curious to get your reaction to them. One anecdote is a university president in the United States, who's both American presidents, who said, yeah, so chat GPT, huge disruptor. We have to go 100%.
[0:15:48] Back to oral exams. No written anything. We've got to go back to oral exams. The only way to ensure the students are being trained. So that was one side of the spectrum. Another side of the spectrum was another university president, very honest, really like him. And he said, look, we just need to admit that we have no idea what the students are learning. So we have to get out of the role of certifier.
[0:16:17] And just tell an employer, look, these students spent four years here. Here are the courses they signed up for. We don't know if they showed up for class. We don't know if they learned anything. We don't know if they actually took the test. But you go figure it out now as the employer. We just gave them the opportunity. So, of course, we have a very radically different approach than those two. But when you hear kind of reactions like this, how do you process that? I think this is the normal.
[0:16:46] Range of human reactions to going through a moment like we're going through. It seems intuitively quite wrong to me to say we're just going to go to oral exams. It also seems intuitively wrong to me to say you have to make up your own mind entirely, but maybe it isn't because what an employer cares about is what someone can get done. And if these tools have so changed what they can get done that the educational process
[0:17:18] I mean, I have a lot of sympathy to that. Yeah. No, I couldn't...
[0:17:47] I couldn't agree more. I think it really is interesting to see how these institutions are thinking about evolving and how some of them are resisting it. And I do think that we are going to see big, big reform movements. One very prominent example I know that we've been involved with is there's a university in the United Arab Emirates. There are only three federal universities in the country. And one of them, Zayed University,
[0:18:15] was led by a really visionary president, Her Excellency Noor El-Kabi, and she realized that the university, named after the founder of the country, didn't live up to the namesake that really needed to up its game. And before ChatGPT, before OpenAI, or before OpenAI released, of course, to the general public, after OpenAI existed, of course, she said, no, we've got to up our game.
[0:18:44] And she initially had us build a college of interdisciplinary studies within the universities. The university educates 10,000 undergraduates and thousands of graduate students. And almost instantly, she and the board decided to shift the entire incoming class into the new college and effectively to sunset the old programs. And the howls of protests.
[0:19:13] That came out of that decision were deafening. And then, I have you to thank for this, when ChatGPT came out and Zayed was the only fully interdisciplinary, fully post AI ready institution in the entire region, if not the world, all of a sudden, all the criticism went immediately silent. And so that's where I think bold leadership is going to come because institutions are going to be sclerotic.
[0:19:43] Unless they have a leader that's going to push them in that way. You just spent a month, right, traveling the world, meeting various leaders in government and civic society. I mean, did you get a sense that people are ready to do the things or are they just kind of looking at it and saying, oh, wow, that's scary and I'll let my successor figure this out? Some of both, of course, but on the whole, like,
[0:20:14] I was very impressed by how much people want to engage with this, are taking it really seriously, are up to speed, understand the magnitude, and want to kind of leave their mark and contribute. And do you think that there's a hunger to take on the entrenched systems to change them? Because ultimately that's where push comes to shove, right? I do. I really do. I was very impressed. Wow.
[0:20:44] If you had to give the people who are here watching, listening, a tip of the hat to where they should be looking for, what are the societies, the leaders that you think are going to make some of the, and again, it's a prediction, you don't know if they'll actually do it, but again, if you were just wagering, where do you think some of those system changes are going to come up first?
[0:21:13] I think we're seeing it the most right now in creative work. And not necessarily, I mean, I'm sure in some ways, but not in the main way that people were really terrified about when the first image generators launched, which is there's going to be no role for artists or visual artists or whatever. What we're seeing is that what a visual artist can do now has just gone to a totally new level. Now, I don't want to minimize also replacing.
[0:21:41] Some people, because that's definitely happening and very painful. But the quality of creative work, I think, is going to go up so much, which sounds really good. But imagine if TikTok gets 10 times better and 10 times more addictive. Maybe that's not so good. But I think that's going to be a big change coming soon. Fascinating. And do you see that embraced globally?
[0:22:08] Do you see that that's just going to be a sector-wide phenomenon? Because when you have the massively more productive people, there's just going to be nowhere for the non-productive ones to hide? There, I don't really know how that's going to play out. But I think that, I mean, I think everybody will get more productive. There probably will be more dispersion, more spread. But I think quality of life and ability to contribute should go up across the board. But as you said, maybe unevenly.
[0:22:39] Yeah, you know, it feels to me like the uneven spread of the spoils, economic spoils of using these tools is in some ways you could think, well, you know, if you've got a 100x producer over here and a 1x producer over there, there's going to be even greater economic disparities. At the same time, one of the...
[0:23:08] The core problems in society, certainly something that we've tried to tackle at Minerva through our own university, through what we're talking about broadly, is to realize that talent is dramatically more broadly distributed than in a traditional sense that you would look at in traditionally highly selective institutions. So as an example, historically at Minerva University, we've had five
[0:23:38] And the outcomes of Minerva students has by pretty much any measure, employment, entrepreneurship, even graduate school placement, which is perhaps the most remarkable given the entrenched systems,
[0:24:08] has outpaced the graduates that come from huge privileged backgrounds, right? And so to me, I think of ChatGPT as an accelerator of that, right? Because the guy who's going to go and say, oh, you know, I'm going to hire my golfing buddy's son, right? And yeah, you know, he's not as good as the best person I can interview, but okay, whatever. I'll make a 20% sacrifice in productivity and the guy will owe me a favor. Now,
[0:24:37] It could be a 99% hit on productivity, right? And so all of a sudden, you're not going to... I mean, maybe you're going to have dramatically different economic value for the people who can produce at that level, but at least my hope, at least if I were to be able to shape the future more and more, that economic value is going to...
[0:25:03] to accrue to the people who have merit as opposed to the people who happen to be born into a wealthy family that will then perpetuate that. And this to me is the crux of those educational institutions that will rise to the challenge, right? Because those that are going to find and nurture talent that will teach them how to be wise in order to be able to leverage these tools, their graduates are going to crush.
[0:25:33] I still hesitate to call it an equalizing force, but it is like a force in favor of better justice or something like that. I love that.
[0:26:00] Although, we may find out that just because someone is smart does not mean they're supposed to do better in the world, which is certainly the case right now in justice for all those things. Right, right. Well, that's obviously true. But talk more a little bit about that, right? Because ultimately, I do think that this technology will change the economic distribution in the world, individually, societally, sector-wise, et cetera.
[0:26:30] There's got to be some profound scenario planning, let's put it that way. Rather than predicting the future, thinking about scenarios that could come to pass and then how do we mitigate or try and mitigate some of the potential downsides. If you had to riff on that, what would you be looking out for about decisions that we make today
[0:26:58] From a policy perspective or from a values perspective or promoting a particular approach, that could help mitigate some of these pitfalls in the future. I think it's good to study a lot of things today, but I'm hesitant to recommend too much today. Everybody wants to pontificate on here's the master plan all the way through. We're going to go where the technology takes us to some degree. Society does get a lot of influence, but there are things like...
[0:27:27] The intersection of society and technology, what technology can do, what society wants, it's just not so obvious always. And so I think, I don't think like UBI, Universal Basic Income, I don't think that's a full solution, certainly not. But I think it's part of it, or maybe part of it. We should study it. We should study other ideas like it. One thing that I would say with confidence, there's really not much, but one thing I would say with confidence is that
[0:27:56] People really want agency over their future, our future. And that's super understandable. And I think one mistake a lot of people in the AI field are making is to just say, well, you know, there's going to be no jobs, but everybody will be rich and happy. And that sounds so dystopic. I don't think it's what's likely to happen at all, by the way. As we've been saying, I think people will just find new and more amazing things to do and human desire.
[0:28:26] And creativity and status games and whatever is going to turn out to be pretty limitless. But I think this is a good time to study lots of ideas and be very open-minded about what the eventual solution is. Yeah. Yeah, I think that's such an important thing. And I can't tell you how many times I've been in events and people are like...
[0:28:52] Talking with certainty about this is where the future is going to head and this is the implication and all the rest. Whereas to me, I think the only thing that we can have any level of assuredness on is that as humanity has progressed, we've constantly seen technologies that have allowed us
[0:29:20] to retreat to higher cognitive levels. From the very earliest tools that made physical labor somewhat easier, all the way to the industrial revolution, all the way to the digital revolution, computer revolution, which kind of took as many of the mundane tasks as we can and said, okay, there's more and more value accrued to more complex thinking. But I think...
[0:29:49] What's so profound about this change and this is why I do hesitate to think about the changes that have happened in the past is this does enable a level of cognitive understanding of intelligence to show up.
[0:30:11] In such a way that the nature, we just can't retreat in a linear fashion, right? We can't just say, oh, okay, now all of the grunt work is going to be done and I just can be going higher and higher. The reality is that human beings need systems of thinking, need frameworks to be able to apply to a different kind of system that gives
[0:30:39] solutions and answers that actually creates product that isn't just providing information or automating labor or things like that. So to me, one of the great fears that I have is that unlike these tools that have happened in the past, where the tool would come up,
[0:31:03] The washing machine and the computer and the internet and basic fundamental technologies, we were able to stay ahead of them. We were able to say, okay, that intersection of technology and society, yes, society will leverage that. And I think now with disinformation, with political divisions, with this algorithmic spread of hatred and othering and everything that gets involved,
[0:31:32] The technologies are outpacing our ability to deal with them as humans. And I think that the fundamental problem is that we are not systematic thinkers naturally. Human beings can be, but we need to be trained to be systematic thinkers. And if we're not trained to be systematic thinkers, the technology will outpace us.
[0:32:01] Again, who knows what will happen, but I cannot imagine that it's going to be a good thing. And so for me, the urgency of getting the educational system to work and get people the tools they need to navigate this place, I don't think there's anything more important right now. And to your point, the innovation imperative, the need for institutions to
[0:32:30] Amen.
[0:33:00] That there are going to be individuals, institutions that are going to experiment and their experiments turn out right, right? In the sense that their graduates, those individuals are going to be better prepared for the world. They're going to be ready to deal with complex challenges, etc. What do you think are going to be the mechanisms, the systems, that intersection of technology and society
[0:33:30] that are going to best leverage those individuals, that are going to best reward those institutions, as opposed to having, you know, I often reference Idiocracy, the movie, as a wonderful future documentary, as opposed to just creating the idiocracy, as opposed to just, you know, kind of going back to our first kind of premise. How do we reward quality?
[0:33:59] As opposed to mediocrity, how do we get hard work and inventiveness to triumph over sloth and laziness? What are some of those mechanisms? I really believe that just is going to happen naturally. And the more people have leverage, the more that is going to be rewarded. Yeah. So you really believe that the market will ultimately reward that?
[0:34:28] I do. Yeah, I do. That's one area where the optimist in me, I really profoundly hope that it will. And the pessimist in me has a note of caution given what I see happening in the world from previous areas. And that is one area where I'm...
[0:34:58] I know, particularly cautious. We only have a couple minutes left. I wonder, in this context, for people that are listening to us now, that may be listening to us in the future, those that are no longer in the time of their lives when they were intensely studying. Of course, we're all learning, we're all changing. But people who are making the decisions out there,
[0:35:30] Two or three pieces of advice. What can they do? What can they actualize to get not only the organizations they lead, but the societies they care about better prepared? What would you think they should be doing? Well, I think nothing beats direct experience and fluency with the tools.
[0:36:00] The data that I've seen, the people who actually are using ChatGPT and services built on things like the OpenAI API every day have much better intuitions about what to do. So I think that's my top piece of advice. It's just like, and I think this is true for most things, but the closer you are to it as an actual user, the better the quality of your intuition and such will be. Amazing. I think that's so important.
[0:36:28] Our mutual friend, Reid Hoffman, who's, I think, one of the most amazing positive forces for AI. I know you and I both did his most recent podcast, Possible. He always likes to think in the optimistic way, what it can unleash and unfold. Again, without too much prediction, but actually talking about what you have seen.
[0:36:58] Maybe it's not fully rolled out. What's possible? What is the thing that has, or what are some things that have just blown you away because you've seen it work and most people haven't yet? Well, we try not to talk about unreleased products, but with GPT-4, I mean, education, healthcare, personal productivity, those are all pretty big. Yeah.
[0:37:29] What about from an entrepreneurial perspective? For example, I haven't played with it yet, but I just saw a new company launched, Omniscient. Have you seen this one? Where you provide a prompt, and in eight minutes, Omniscient, I should say, not Omniscient, Omniscient. In eight minutes, it will produce either a fiction or a non-fiction book.
[0:38:00] I imagine the first version is not going to be ready to be winning a Pulitzer, but imagine if you could just feed data and in eight minutes have a draft of a book, fiction or nonfiction, and then work off of that. To me, that is a mind-blowing application.
[0:38:25] That I don't think most people know about. Any that come to your mind in that vein of people that are out there using the tool in profound ways? I mean, another version related to that is the number of people who come up to me and say, hey, you know, I like ChatGPT, but my kid loves ChatGPT. And every night we're like telling new creative bedtime stories. And it's, you know, just this beautiful creative force to see unleashed. That's very cool. That's awesome. Great. Sam, awesome chatting with you.
[0:38:56] Thank you for the time. Thank you. Thank you, everybody.