Welcome to the stage co-founder and managing partner of 01 Advisors, Dick Costolo, and founder, serial entrepreneur, investor, and creator, Brad Keywell.
I feel like doing a fireside chat in the opera house is like underwhelming. One of us should do like a pooini area or something. Vincero was pretty good. I'm not going to take the same risk as Dick just took.
I sound great from up there.
You really do. It's a good vibration.
Okay. We're here to talk about something that's really not that big of a deal and really doesn't matter that much. AI. So this is probably the most important thing that's ever happened in our lifetimes. And I want to start by framing what it is at a remedial sense. I think it's important to understand really what's happening because philosophically all of technology has been built off of the idea and dependent on the idea of a database and a database is organized in rows and columns and in each row and column is a piece of data and there's been a way to tag that data and all of the innovation before large language models and the like have been dependent upon innovative ways to search the rows and the columns and the tags. End of story. So Oracle is a way to search rows, columns and tags, etc., etc. And fundamentally what's happened just to level set this is an amazing idea came about taking every word that's in every row and every column and making it dynamic. So every word can learn from every other word. And that's what's happening right now is we're witnessing the innovation that comes from that idea. So that's one way to look at framing the picture and I go to you Dick with this framework we're going to have today. So I'm going to interview Dick and we're just going to talk about stuff but I'll keep it going. But the question is what is the big picture happening right now from your perspective?
Yeah, I mean like you said it all this all started in 2017 really with the paper Attention Is All You Need which introduced the transformer model that all these LLMs are based on and that really created this quantum leap and speed of execution toward artificial general intelligence. The big picture is we're probably depending on who you want to ask three to five years away from true artificial general intelligence. AGI being the machine is just as smart as or smarter as humans on any particular question, text, reasoning, etc. And they're probably not that far away after that from artificial super intelligence. So the really big picture is you've got companies in the US and organizations/companies in China racing each other toward AGI with OpenAI probably being the furthest ahead in the US. Although these models continuously displace each other as to who's in the lead and you know DeepMind of course has been in the news from China.
And how do you think about speed? Because I think you and I have I've known you for 20 years, right? We've been in the trenches and you witness speed and you can't believe it and all of a sudden you can't believe that you couldn't believe it. I mean what's going on now?
Speed is probably the you know people always talk about technology moats and investing and what's their moat? What's the thing that prevents someone from catching up to them? In the AI space, speed is the only moat. It really is. And so the interesting challenge and I guess this is all framework for thinking about how to invest in the space and then also thinking about the good and bad things that could happen to society from all this is one of the core parts of AI theory if you will is alignment. And alignment is let's make sure the AI is doing things that are aligned with human intentions and beliefs and goals, right? And so let's make sure the AI doesn't get out of the guardrails and do something it shouldn't have done that's not aligned with the way we think about our goals as humans. Well, the big challenge is in the race to win and winning is defined as well we have to beat China and then China they're like well we have to beat the US. Some of the alignment issues are falling by the wayside in an effort to win and be first because if we don't get there and they beat us they can take out our power grid and on and on and on. And I think there are enormous implications for getting to AGI and not having alignment and you know there with all sorts of catastrophic consequences as well but speed is and the train has left the station on that. You can there are companies that are like well OpenAI needs to slow down and obviously a bunch of executives have left the company and saying you know we're not focused enough on safety but the train has really left the station on that front and it's all about getting there first.
And I want to get to alignment when we talk about the human consequences of what's going on. I want to first start with pattern recognition. Okay. Because you and I have both been starting companies and investing in companies for 30 years. Social media had a purpose and then it had a consequence.
So I'd say the purpose was to connect us. The consequence...
Yeah. Well, I'm smiling because a lot of the and the reason I mentioned alignment earlier is a lot of the biggest proponents of AI, I mean Reid Hoffman, who's a friend of mine and the founder of LinkedIn, of course, Reid will get on stage and talk about how great AI is and you know, you really shouldn't worry about your kids going and doing their homework on it because it's just going to help them learn faster. And when I talked to him after one of his conferences and I was like dude were you around for what just happened in social media? We had 2009, 2010 we were connecting the world we're giving the smallest voices a voice in the world and by 2016 we were oh actually people are using it to try to control our elections and outcomes of elections and misinformation and abuse and causing people to commit suicide because of being abused on social media. So the implications and consequences of technology are not always and or almost rarely ever only good.
Okay. So let's go back even further. The purpose of technology in general is to make us humans better in some way.
I suppose there is some technology whose purpose is to destroy and ruin. Let's just stick with for this conversation because we're investors. So if the purpose we're optimists as investors of course and what we're investing in is...
You could be both an optimist and a realist, you know.
Yeah of course. So but technology in general the purpose has been to advance humanity. The consequence generally speaking what you say is what of technology?
I think the consequence has been a bit of a dumbing down of all humans. So I'm going to go get scientific for two seconds. All humans let's say we burn if we're not working out we burn like whatever 2,000 calories a day 30% of that is just your brain. So literally 30% of the calories you burn sitting around all day are just your brain so evolutionarily we're trained to do cognitive offloading meaning I got to get something else to do some of the work my brain would otherwise be doing because I'm taking up too much energy literally. So using a calculator is cognitive offloading right? I don't want to calculate 20% of the bill I'm just going to type it into the calculator and do it. And you see and that causes people to be less and less capable at doing simple things like calculating 20% of a bill. Well to the extent you can now go to ChatGPT and cognitive offload everything...
So it's got a tremendous dumbing. So the purpose of AI is again to make us better.
And it will make us way more productive. It's going to make us a lot more productive. We're just going to be reliant on it.
So what I think is when I think about the meta question, I think about what it takes for a human being to experience something making their lives easier, right? And what we're used to is some time to think, to resist, to bureaucratize as a way of resistance. An organization can bureaucratize itself to slow down technology's pace. Right here we are where the speed is such that I think all of us are feeling a bit of overwhelm because we're still human. How do you think about that problem or that reality?
It's a great question. I'll actually address the bureaucratization piece first in the interest and thesis around investing in the space. I think traditionally when you look at a new wave of technology you look at the incumbents in technology and you think well they've got leverage in the space and they're going to be able to out innovate. And since AI is so specifically expensive you've got to spend money on compute and it's quite expensive. You would think that the Apples and the Googles of the world would be in the leadership position to go affect change here. Particularly because the authors of the Transformer paper were Google employees. And yet Apple and Google I like to say they ship their org chart like the products like you can see the org chart in the product which is a very bad thing. And so as a result, you've got these AI first companies that have emerged that are just much much faster in all respects. Not just OpenAI versus Google for example, but you think about things like Siri on iOS, it's useless compared to any of the emergent voice-to-text apps like voice notes or Granola the note-taker, etc. So this will be a space where the Gen AI from the bottom up companies from the ground up as Gen AI will be able to move a lot more quickly and innovate more rapidly and win versus the incumbents not just in the general tech space like Apple and Google versus the OpenAIs of the world but in industry verticals. So if you think about accounting firms or legal law firms or even companies like Palantir, okay, Palantir was started 15 years ago and it wasn't Gen AI based. There are emergent companies now that are AI first with very small teams that are taking on Palantir and are incredibly well funded and I think that you'll see more and more of that industries being entirely reframed by companies that are created from the ground up. Gen AI.
I think what's interesting philosophically is this is probably the first time where we can't escape that technology is happening to us. When there's bureaucracy in organizations, you can get the feeling even though it might not be true that it's happening by you because you can have some agency in pacing it and slowing it down, right? Because when I built companies in the enterprise space, the people that work in the companies that we are quote unquote disrupting are the barrier. And I started to realize it's because they're exercising their agency as humans, which is I'll tell you when we're ready to change. I'll tell you where we're going to change first. And now we're in a position or in a state of affairs where for all of us people whether we work in a company or whether we're side hustlers or whatever it is technology is happening to us and I don't think there's a framework for us to think about that.
No I agree with that and as and again as you think about again I'll talk about maybe I'll talk a little bit about investing and then a little bit about the potentially negative consequences because I don't think we've done that enough in technology. Every industry vertical that you can imagine is going to be re-imagined from Gen AI from the ground up. So right now there are multiple venture firms who are investing in going out and buying small to medium-sized accounting firms and they're going to replace the accountants with AI. Now today the LLMs are not very the models are not great at math but these venture firms are knowing they will become great at math are getting ahead of that and they're buying these firms and they're going to replace the people with AI. That's going to happen in legal they're going to be ground-up law firms that's going to happen in healthcare.
General Catalyst bought a hospital system to own a hospital system to then replace the back end of it right so the consequence is going to be that I think that people and government aren't so there's an investment opportunity. You can pick any industry vertical and go attack that. Redo it from the ground up with Gen AI with much far fewer people so you can charge a lot less and still make greater margins. And that will work in every industry.
The consequence is going to be you're going to have an evisceration of the white collar class. And I don't think people are talking about that enough. That's going to lead to an enormous sense of fear. Fear of safety, fear of my job. It's going to lead to political radicalization because you'll end up having the very very wealthy and when you get rid of the rungs on the ladder of the corporate ladder and you don't have a I can get this and then promoted to that and then promoted to president, there's just the CEO and a few people and not many others. You're going to have political radicalization. And I don't think that politicians in the country or the technologists are talking about this enough.
And therefore what do you have any ideas about that? Because I look at it and I say I'm not sure there's any clear idea yet. Now granted every time like when I read history the stirrup on the saddle on the horse was radical technology that enabled warfare because it gave humans leverage to ride a horse. So start there. Every ounce of technology development that's happened, humans have found a way to not be eviscerated, to not be sitting looking at a wall all day, right? And so now I suppose it's just amplified. So what do you see happening?
I think that it's incumbent upon the industry and I mean the AI industry and the leaders in AI in the industry and Dario from Anthropic has been to his credit has been super vocal about this. Is like, hey, we're going to lose millions of jobs. We need to be talking about what to do here. And I think it needs to lead to something like universal basic income where there's just the government is paying people even though they're not working and the country at large is far more productive. Like GDP could go up with mass but also have mass unemployment. And so I think that's a core component to investing in this space ethically. I think is going to have to become some of these firms that are thinking about what are we going to be able to give back to the government to take care of these people who are losing their jobs. The challenge in the United States is it's a particularly bad country for something like UBI because or poorly aligned country if you will for UBI because you've got both a free markets above all else way of thinking about capitalism which has gotten us this far and it's been great but also just an immense general distrust of the government being in charge of stuff.
Yeah. I was talking to Dan Gilbert yesterday about the collaboration of private industry and government and the role of an entrepreneur in a city. And here we're talking about the role of entrepreneur in society. And that's not created yet. This is a new paradigm. We're going to have we collectively are going to have to figure out how it is that some of these entrepreneurial leaders or someone collaborates with government because government alone I don't believe is going to be able to figure this out on itself.
I don't know how it's going to happen. I believe that it has to happen.
You're 100% correct. And almost always the technologists in charge at the time are not ready for the obligation that's in front of them. I'll give you an example from Twitter. So I'm running Twitter and my assistant comes into the office one day when I have a meeting with my staff and says like, hey, the she's whispering. I remember she's whispering and then later on I was like why are you whispering that anyway? She goes hey the White House is on the phone and I was like how do you know like how do you know anyone can call and say they're the White House? She's like no I think it's Valerie Jarrett Obama's chief of staff. And I was like I mean again how do you like she just anyone can call. Anyway I answer the phone. I'm like you know and I like born and raised in Detroit. I got the Midwest upbringing. I'm like running this tech company. I'm like what did I do? I was like I didn't do anything. What do they want me for? And I got I literally my first thought was I didn't do anything. And I got on the phone and I'm like, hello? And she's like, hey, the president would love to come and have you come and talk to him this week in the Roosevelt room about misinformation. And Snowden had just leaked all that information and we need to understand how terrorists are using your platform. Like, can you be in the Roosevelt room on Friday? And I was like, yes. You know like so but again even though this stuff was happening on the platform the people in charge of these companies present company included you're not you're like...
That yeah but in the AI world these all of these entrepreneurs and technologists and again to Dario's credit he is but I don't think they I don't think any of them are enough have to be focused on this or it's going to be it's going to be horrible.
Yeah. If anything all of our obligations whether it's founder part of a company as a team member, investor, board member, everything is magnified. Everyone like what we do matters more than ever. I want to stick on Twitter for a second and keep jumping around and go back to the idea of why is it that an Apple is not dominant in AI? And what I'm seeing is the capacity of one, three, five, maybe 20 people to fundamentally out compete and out create a 2,000 person company. And it goes back to my core idea, which I often talk about. I'm like, one person with a great idea is unbelievable. Two, it's a little political trying to navigate between those two people. Who how do you figure out what idea you're going to work on? Three, you're totally screwed. The politics of a meeting of three people, how are you going to make a decision? Gets hard and keep adding exponentially. So, 2,000 people, wow, is that hard to navigate. 10, 2, 50, a lot easier. And here we are watching Apple be outflanked. How do you think about that?
Well, you're absolutely right. I mean there's a reason Apple was so successful when Steve Jobs returned because he just told people what you know he just he decided he would go around the room and ask and then would say you know do this. I mean Eddie Cue famously tells the story of when he first met Steve Jobs he goes they were working on the iPod and they're in a marketing meeting and this is actually it actually interestingly really informs how to think about again investing in the space. In this meeting Eddie Cue who runs all marketing for Apple and did under Steve as well. In his first meeting ever with Steve, he's in the room and they're talking about marketing the iPod and the CMO is presenting something to Steve and Eddie says something in the meeting that disagrees with what the speaker is saying and Steve looks at him and goes, hey, shut up. And then the speaker keeps talking and 20 minutes later she comes to some conclusion that Steve realizes or sorry that Eddie realizes as well it's based on this wrong slide she's got earlier in the deck and he says you know yeah but slide 10 is wrong so this is really you know and Steve looks at him and goes hey I thought I told you to shut the [__] up. And so you had a company where the guy in charge may be crazy but he's a lone voice and in charge and he's going to do things the way he wants to do them. Of course, the way that meeting finally ends is Eddie says something at the end of the meeting that is contrarian but agrees with Steve against the opposition of everyone else in the room and Steve goes finally someone in this room thinking for themselves anyway. But so but to your point these AI companies that have a visionary founder that doesn't have to deal with any legacy technology and can use these models that are already capable of writing 90% of the code you need to go forward are going to beat any incumbent in the space irrespective of how good that incumbent's technology team is because they've got existing technology which is baggage and they have to support that while trying to move into this new age.
Yeah, we're talking about a friend of ours runs Uber and you know two years ago 30 last year 30% less coders this year I think order of magnitude is much greater than that. So it's happening before our eyes. The descaling of some of these large organizations attempting to create agility where there wasn't one before. I'm going to get to investing in one more second. US versus China. It's worth talking about for a couple minutes. So is this peacetime creation of AI? Is this a peacetime race or is this a wartime race? Is AI being created to help us be more efficient or is it really at fundamental core right now a tool that is of existential concern which might explain why it's existentially important to each country? How do you think about it?
I mean the short answer is it's both. But I think if you really the more complicated answer is it's 100% a race to be first because the sovereign defense of the country is going to benefit the winner significantly. Like if you lose and the other AI can take out your electricity grid or take down your AI and you can imagine how the compounding problems that occur there which is again goes back to means that both the Chinese and the American companies are racing to do this without considering alignment enough which is you know again...
