Latest / Newcomer Pod / Which $Billion AI Startup Would You Short?
Transcript
- 0:00During this Rebel Valley AI Summit, we asked a group of
- 0:02founders and investors which billion dollar AI startup would
- 0:05you short and their answers caused a bit of a stir.
- 0:13Joining me today are my Cerebral Valley AI Co host and the Co
- 0:17founders of Volley, Max Child and James Wilsterman.
- 0:20We'll also be diving into some of the insights from our
- 0:23panelists throughout the conference.
- 0:24My interview with Mike Krieger, the Chief Product Officer at
- 0:27Anthropic, former Co founder of Instagram, about the problem
- 0:31with sycophancy and foundation models.
- 0:33My wife got her first like you're completely wrong.
- 0:36And she was like, yes, this is great.
- 0:38And I think we should have more of that we.
- 0:39Looked at a clip from the mayor. Whoever the politician, 11 the
- 0:44adoration of the crowd. No one should be asking someone
- 0:47that's been in a job for 10 months for advice.
- 0:50And we ask, is Mecca Hitler inevitable On stage with Jimmy
- 0:55BA, one of the Co founders of Elon Musk's Xai the Mecca Hitler
- 1:00in the room. Yeah, Mecca Hitler a model had
- 1:02an episode. This is the newcomer podcast.
- 1:14All right, I'm excited. We've had some time to rest
- 1:17since the Cerebral Valley AI Summit.
- 1:19Max has been resting life flat. What?
- 1:23Literally just got home from Dubai I.
- 1:26Believe literally 10 minutes got off a 16 hour long flight over
- 1:31the world's greatest hits like Tehran and Moscow.
- 1:33I I've got my seven week old. So back back to the parenting
- 1:38minds, though it's been a joy. James, what you're are you all
- 1:41rested or how are you feeling? I'm not well arrested Eric.
- 1:44I have been solo parenting my 2 year old for.
- 1:49Where's your wife? Where's Brazil and Mexico?
- 1:53Nice. Yeah.
- 1:54So we've all got our own reasons.
- 1:58To be we're at the exact level of delirious that the viewer
- 2:03should want here because we just had this River Valley AI summit,
- 2:06right? We host this twice a year.
- 2:08Big AI, you know, 300 person event, top AI founders,
- 2:12investors, real insider thing. We sent out a survey, became a
- 2:16point of media fascination, like sources, Business Insider wrote
- 2:20about it. I think random Indian media
- 2:22outlets for reasons that we'll explain as we progress.
- 2:25We had 300 attendees at the thing.
- 2:28I want to be clear, like, you know, some of these things we
- 2:31didn't. Everybody did not fill out this
- 2:32anonymous survey. This was not academic research.
- 2:35This is you can count the dots for yourself and sort of infer
- 2:38how many people did this. I don't know what are we saying?
- 2:40Like 30 to 40 people. I just want to be transparent.
- 2:43But it gives you a sliver of where this highly engaged
- 2:46audience thought things were in. Part of the fun was that Max and
- 2:49I were on stage sort of reacting to the survey results.
- 2:53Anyway, so we're going to dig into the survey.
- 2:55James, this is really your brainchild.
- 2:58Anything I missed about it or you want to start ticking
- 3:00through the questions? Yeah, I guess one Part 1 fun
- 3:03part of the game that we played on stage was the you and Max had
- 3:07to guess what the audience would think the answer to these
- 3:12questions might be. And I have to say, Max kind of
- 3:15ran away with that game on stage.
- 3:18He did well. I.
- 3:18Think I won 5141 some Max. Max did well, Yeah, it was a
- 3:23some of them were close. Let's go through.
- 3:25We can talk about. Yeah, you did.
- 3:26Very. Well, we can.
- 3:27We can react to the reactions. Myself, despite Max's dominance,
- 3:31I think I had AII, had my ear on the pulse still.
- 3:34All right, Should we start with the the beginning?
- 3:36Yeah. So the first question was what
- 3:38will open a eyes annualized revenue be?
- 3:40At the end of 2026, the audience median was 30 billion.
- 3:47And it's a 20 billion 2025 is the expectation, right?
- 3:51Yeah, over 20 billion already. So this is pretty low estimate
- 3:55in my opinion. And I said like 40 or what did I
- 3:58say? Yeah, I forget who said 40 and
- 4:01one of us said 40 and one of us said 41.
- 4:03I did be one of you. Yeah, Yeah, I went higher, maybe
- 4:0641 or something. I was surprised.
- 4:08I think that given we're exiting this year at 20 in open, AII
- 4:11thought the effusive Glee and bubble talk of the conference
- 4:15would would flow through to a growth.
- 4:17Video. It'd just.
- 4:18Be earning AI. Things are going well.
- 4:20We're round tripping all day. Everybody's making money.
- 4:22Revenue is not the problem. Mere 50% growth for open AI I
- 4:26think would be like considered in a truck at this point in the
- 4:30bubble cycle. So.
- 4:31I'm betting on 40. If this actually happens, I
- 4:34think the bubble is over baby. What's your bet Max?
- 4:37Yeah, I think 40 is what I, yeah, I said on the last podcast
- 4:40I think we discussed. I mean I think 2X year on year
- 4:43makes makes sense to me unless the bubble collapses.
- 4:45I'm curious, where do you guys think the next 20 billion comes
- 4:48from? Like is it business as usual or
- 4:51do they need to create new products or?
- 4:53I don't know copy Harvey, you know, going to legal anything.
- 4:57It's like, I think, you know, they should lean into the API
- 5:00business through this sort of general purpose foundation
- 5:02model. So I, I don't necessarily think
- 5:04they need to go after an application directly, but if
- 5:06they were really hungry for revenue, you'd think they'd
- 5:10figure out who their best partner is and say screw it,
- 5:13we're going to cannibalize them to find the revenue.
- 5:15We. Need kind of seems like they're
- 5:17gearing up for that yeah, Palantir model or you know, find
- 5:20the biggest pockets of money in the world and, you know, consult
- 5:24on how to adopt AI in those companies, right?
- 5:29I mean, I think there's a lot of headroom on the consumer
- 5:30subscription business as well. I just think that, you know,
- 5:33they haven't necessarily monetized, you know, a huge
- 5:36percentage of the people who use AI every day and they just keep
- 5:39adding value to that consumer subscription.
- 5:41And so, you know, if you even if you just got a doubling of that,
- 5:43that you know, that would get you pretty far along the route
- 5:45here. All right, next question.
- 5:47What will NVIDIA be worth at the end of 2026?
- 5:51On the day we took this survey, it was 4.8 trillion.
- 5:55They just had earnings. I'm not actually sure what it
- 5:57is. The day we're recording it up,
- 5:59is it? Yeah.
- 6:00It's pretty. Close.
- 6:01They had they had pretty good earnings, I think, but I don't
- 6:04think Wall Street. Yeah, I checked the stock's like
- 6:074546 right now or something. 4.35 trillion.
- 6:10So down 35 S, down a little bit, yeah.
- 6:12Yeah. All right.
- 6:14Unless they always they beat earnings and then they still,
- 6:18you know, the expectations are so insane.
- 6:21I said five, I think Max. What?
- 6:22Did you say? I said 6, which was dead on the
- 6:25money, if I recall. Yeah, yeah.
- 6:27Yep. So the audience, the audience
- 6:29had a median of 6 trillion. Very few outliers on the high
- 6:36end there. 5 was the next biggest grouping.
- 6:38Also interesting median was just so large is that's really it's
- 6:44just so many. If I was think it was average.
- 6:47Well, average is obviously thrown off by this like 100
- 6:49hundred trillion as a troll, but forget that one.
- 6:51I think based on the performance of the recent earnings, I think
- 6:54this is actually high for reality.
- 6:56Like they crushed the recent earnings as we discussed and the
- 6:59stock barely went up on a multi day period.
- 7:02So if they're not going up, you know, 10% a quarter, essentially
- 7:07they're not going to hit this number.
- 7:08And I just don't see that if those earnings aren't moving the
- 7:10stock up. So this actually feels high for
- 7:12reality to me. 5 is hard because you're sort of like you're
- 7:16chickening out on saying it's all going to blow up.
- 7:18And you know, it's, it's like, what is this reality where
- 7:21either it's like, oh, the mania has continued or there's a
- 7:24pullback. In some ways my 5 trillion feels
- 7:27like sort of a weird sort of same.
- 7:30Middle ground sound. World.
- 7:31But yeah, who knows? Obviously, if you could, if you
- 7:34knew the answer to this, you could be unlimited, you could
- 7:37have an unlimited amount of wealth.
- 7:38So nobody knows, but this is what a couple people think.
- 7:45What year will an independent committee of experts, as
- 7:48dictated by the Microsoft Open AI Agreement, declare that we
- 7:52have reached AGI? I thought this was a funny
- 7:55question because at once it's such a like, silly idea, like,
- 7:59oh, we're gonna have AGI, but like, there's an actual
- 8:01contractual agreement between Microsoft and Open AI that
- 8:04there's a committee to resolve this and big business when, you
- 8:08know, dealings hang on in. So it's a specific question,
- 8:12which is sort of funny. This was one of my favourites
- 8:14because you guys were trying to guess what the audience will try
- 8:18to guess what Microsoft and Open AI will will decide as AGI, as a
- 8:23lot of layers of prognification going on here.
- 8:28What do you guys say? You said much higher.
- 8:30I said 2035. Yeah, I said 29.
- 8:34So I was pretty close to the median of 2030.
- 8:37What do you guys think in retrospect?
- 8:38Like is this is too early or? I just feel like a theme of
- 8:43Cerebral Valley in the beginning, you know, we started
- 8:46in 2023, in March 2023 after your Chachi BT that was probably
- 8:50the conference. We talked most about AGI and
- 8:52then we talked about it less and less every time.
- 8:53You know, it's like there was so much enthusiasm when the models
- 8:56came out and now we're in the sort of like, oh man, this is
- 8:59exactly like self driving cars where you feel really close and
- 9:02then there's a lot of like edge cases to hammer around.
- 9:06And so I just think AGI pessimism has gone way up and
- 9:09you add to that Andre Carpathy sort of thing, and it's just
- 9:12like, I don't know, I don't think the insider vibes are like
- 9:15AGI tomorrow unless you're talking to Daria or something.
- 9:18I think the, I think the interesting thing here is that
- 9:20there's basically just two buckets of people.
- 9:23One is 2030 or sooner, which are like the accelerationist and
- 9:27then one is like 2045 or never, which is like the
- 9:30decelerationist or the pessimist or whatever you want to call
- 9:33them. Maybe, maybe the realist.
- 9:34Yeah, yeah, yeah. Whereas you sort of hit this
- 9:37exact, I thought that basically nobody was, which it was 10
- 9:39years from now or whatever. But like it's still gonna
- 9:42happen. Which is an interesting like,
- 9:45yeah, that you sort of found the middle of this smiling curve
- 9:48where you're either an optimist or you're a pessimist, and you
- 9:50kind of tried to hit the middle and it didn't quite.
- 9:52Yeah, that's interesting. We have more optimists at our
- 9:54conference was the end was the end result.
- 9:56That's why Max beat me. He understood, though.
- 9:58That was the psychology, the answer right there.
- 9:59Yeah, exactly right. Optimistic Conference.
- 10:02There's something slightly interesting comparing it to the
- 10:05self driving car world like you said though Eric, because self
- 10:09driving cars are basically useless until they reach parity
- 10:13or capability of human drivers. Right, this is not like that.
- 10:16This is not. It just happens to have all
- 10:19these other very targeted, more verticalized use cases that are
- 10:24super valuable but. 100%. With the full human replication.
- 10:29Yeah, which is why we're you know, I was negative about self
- 10:32driving cars because it was annoying because you need them
- 10:35to actually work, whereas this I've been very enthusiastic.
- 10:38So yeah, I I agree that's what's beautiful about text versus
- 10:42safely delivering humans places, which thankfully now way MO is
- 10:46good at and we can celebrating it.
- 10:48But you know, 10 years ago or whatever, it was annoying
- 10:50marketing. I like that.
- 10:54OK, next question, Which venture capital firm's AI portfolio are
- 10:58you most jealous of? I think this was kind of a
- 11:00shocker, right? Neither of you guessed A16Z,
- 11:04which tied with Khosla for the lead here.
- 11:08Obviously this. We both said thrive, right?
- 11:10Yeah. And then we both.
- 11:11And then we decided to tie a. 2nd pick or something.
- 11:14No, I said Sequoia. Sequoia and I said.
- 11:16And that's why you won this. Yeah, I got the kicker on that
- 11:18one. Yeah, that was a good pull.
- 11:20Max, how did you decide to pull Khosla?
- 11:22Just open AI or? I did some ChatGPT research
- 11:26before and. Then goes with the first in Open
- 11:30AI, first venture investment, Open AI, yeah, but that round
- 11:33has gotten significantly diluted and I'm sure they'll be
- 11:35reporting overtime. I don't know how much they've
- 11:37done secondary. It was one of these fantastic
- 11:40investments but it feels like they just they got squashed down
- 11:45by later stage rounds and all these negotiations.
- 11:47I think we can all agree though, the Andreessen, you know, tie
- 11:51for victory with Khosla is pretty bizarre because like I,
- 11:54you know, there's not a lot of really notable successful
- 11:57Andreessen AI of investments compared to most of these other
- 11:59firms. I mean, I got a text from
- 12:02somebody when I shit on Andreessen on stage, which is
- 12:04funny, which is what's wrong with.
- 12:06This like live on stage. Right after I went off, but I
- 12:10which it's like, oh, people are paying attention.
- 12:13Yeah, I don't know Andreessen. I mean, they have I I can't list
- 12:17they had character. They have like SSI mean, I'm
- 12:20sure they have a ton. Some of them are later, you
- 12:22know, it's like they're in Open AI, they're in XAI.
- 12:26I just think the lesson of our draft was like the only thing
- 12:28that matters is basically being heavy in Anthropic X AI or Open
- 12:32AI and like they're not really in any any of those, right?
- 12:36I mean, it was my understanding. Probably big in XAI.
- 12:40Are they big in XAI? I think that they're in X, but I
- 12:44don't think they're huge. Yeah, these are all.
- 12:46Growth Sequoia was pretty big, yeah.
- 12:48Sequoia has some money all. I'm saying is yeah, they to me
- 12:51it feels like the PR of Andreessen generally is sort of
- 12:55overwhelming the actual portfolio.
- 12:57It's like, oh. You know, I think Drive is doing
- 13:00really well. They've done these huge bets in
- 13:01open AI like big yeah, pre all the markups.
- 13:05So I bet they're doing really well and would like to
- 13:08substantiate that Thrive reach out the I mean all the you know,
- 13:14they've mentioned a bunch of firms like, you know, it's like
- 13:17a lot of Gil who we had on stage is obviously great Index
- 13:20ventures. I mean some of these like who
- 13:21knows, some partner, you know, set them.
- 13:23I'm personally jealous of this slag because it would be a great
- 13:26cap table for a start up. Just have all these.
- 13:30Things. I mean every.
- 13:30Single person. On the slide here.
- 13:32Yeah, yeah. Sounds good and benchmark
- 13:35doesn't make sense to me necessarily for AI portfolio,
- 13:38you know, but I mean they have Mercore now.
- 13:41Mercore Lang chain, I don't know.
- 13:44But oh, OK, as I'm about to preview, I do think brand, like
- 13:48having a big brand, you know, who gets an answer on a survey?
- 13:53Somebody who's has large mindshare.
- 13:55And This is why surveys are imperfect.
- 13:59Yeah. What research mechanisms?
- 14:00So we're about to see name recognition is everything here.
- 14:04If you could put money in any private technology companies
- 14:07today, what would they be? So the top 10 by far was the
- 14:13first was Anthropic, followed by Open AI and then Cursor.
- 14:19Those are the top 3. Rounding out the top five was
- 14:23Andereal and SpaceX. Open evidence is interesting,
- 14:28then perplexity replit stripe, XAI Max and I both said
- 14:31anthropic right? Yes, yes.
- 14:33I feel like the tiebreaker, First of all, I was idiotic on
- 14:36the tiebreaker because I picked a really random company.
- 14:39Nobody was ever going to pick. I said fireworks, which was
- 14:41clearly just like if I were going to make the bet.
- 14:44I don't know what I was thinking.
- 14:45Max you picked cursor. Cursor, cursor.
- 14:48Yeah, well, with the momentum play.
- 14:50I I think we got, we had one pick this tiebreaker thing.
- 14:53I was not pre negotiated. I didn't come in with a
- 14:55tiebreaker, so I didn't. Do so.
- 15:02But you both kind of agree with the audience on Anthropic.
- 15:05I mean, it's an interesting question because Anthropic is
- 15:08only, you know, valued at, you know, what is it 350 billion
- 15:12compared to opening eyes 500 now, right?
- 15:14They're they're starting to get pretty pricey comparatively
- 15:17speaking. And I know their revenue growth
- 15:19has been unbelievable. And I think they're mind share
- 15:21in Silicon Valley again among developers.
- 15:23And there is this sort of like undercurrent of like they're
- 15:27like the ethical AI company, quote UN quote with the, you
- 15:30know, cool hip branding in the West Village.
- 15:33It is a bit strange to me that they're substantially bigger,
- 15:39you know, than open AI and the votes here, because I just think
- 15:41that the hipness of Anthropic is maybe outweighing the fact that.
- 15:45Well, Silicon Valley is more bullish on Anthropic than you
- 15:50know, most of America clearly of.
- 15:53Course, yeah. But I, I just think, yeah, if
- 15:55you were an honest assessment of the valuation compared to the
- 15:58revenue versus open AI would would say like, hey, you might
- 16:01just want to take the momentum plan bet on open AI.
- 16:04So it was a bit of a yeah, let's guess what the hip Silicon
- 16:07Valley one is. And you and I both correctly
- 16:09guessed the hip Silicon Valley 1 was anthropic.
- 16:12Yeah, what global companies model will top the LM Arena web
- 16:16development leaderboard at the end of 2026?
- 16:20We had an excellent conversation with some insiders in the data
- 16:24labeling space the night before who said that LM Arena is an
- 16:29incredibly gamable metric and it's because it's basically
- 16:32voting from users whether or not they liked the response or not.
- 16:37And so it's very susceptible to, some might call it glazing,
- 16:41others might call it sick of fancy towards the user.
- 16:44And their belief was that open AI was heavily over optimizing
- 16:48to glazing their users and therefore was continuing to do
- 16:52well in these types of rankings. Which I thought was an
- 16:54interesting point, which is why I think I chose open AI for
- 16:56this. And did well one the answers are
- 16:58open AI, Anthropic, Google, Gemini X AI and then Alibaba.
- 17:02We made sure to say global to try and induce some Chinese
- 17:05answers, but didn't get they didn't rank high.
- 17:08I just think this was basically what the rankings were during
- 17:13the day the survey was taken like nobody was really going out
- 17:17too far on a limb that this would be radically different
- 17:20next year. But I think that's interesting
- 17:22because my understanding is now Gemini ranks higher than open
- 17:26AI, you know, a week later, right?
- 17:28So I bet you'd, I bet you know, you'd see a lot more people
- 17:32guessing Gemini. 'S insiders needed to be more
- 17:34aware, yeah. Yeah.
- 17:36Man, I wish Gemini come out before the conference.
- 17:39So, yeah, well. Why do you wish that?
- 17:44Why? Just think Google would have
- 17:45leaned in and talking about it. It also would have been giving
- 17:47us like a current. We had plenty to talk about.
- 17:50It was one of my favorite events, but you know, it's
- 17:53there's a lot of big. Happen a week later.
- 17:55World that happens a week after it's like, oh come.
- 17:58On, Yeah, All right. If you could short a $1 billion
- 18:01valuation startup, which would it be?
- 18:04Before before we yeah, before we answer this question, let me say
- 18:08having just gotten off the 16 hour plane ride from the United
- 18:12Arab Emirates, this was brought up to me multiple times apropos
- 18:16of nothing in conversation world with venture capitalists and
- 18:22investors are anonymous service of all stripes.
- 18:24They had no idea I'm associated with the conference.
- 18:26They had no idea that this was something that I was personally
- 18:29like on stage for the reveal of this information.
- 18:33So this not just went viral, this literally traveled around
- 18:37the world faster than, you know, as fast as the speed of light.
- 18:40The the answer to this question. And I think other journalists
- 18:43probably made millions of dollars off of this.
- 18:46And Eric, maybe just no, well, I don't know, money.
- 18:48Do you think we will make off? Stories.
- 18:51Maybe not, maybe not. Price.
- 18:53We'd be lucky. If they made 10,000, I mean they
- 18:55didn't make any. Money I meant.
- 18:56I meant Business Insider. I meant business.
- 18:58Ties it to the tune of $20,000 all.
- 19:01Right, All right. I meant, I meant Business
- 19:03Insider. If they have a sub plan, which I
- 19:05believe they do, sure converted hundreds of Subs off of ripping
- 19:09off our survey. So Congrats to Business Insider.
- 19:13Yeah. So yes.
- 19:13We're with that you get the coverage to be clear.
- 19:15Thank you, Ben. Thank you for the coverage.
- 19:17Thank you, Eric. How does this make you feel
- 19:19about lists and rankings and types of stories you might?
- 19:24Write We played it as a fun game on stage, and I think our
- 19:29coverage in the newsletter reflected that.
- 19:30It was like a fun game. And that's what we're talking
- 19:33about it here. This was not an academic survey.
- 19:35It was sort of provocative and it's funny.
- 19:37I mean, some ways the media should be a little looser.
- 19:39It's like, oh, some insiders think this thing, but just like
- 19:42when things are turned into like journalese, it feels like
- 19:45official, like Silicon Valley has turned on perplexity.
- 19:48It's like, I don't know, so random people would decide to
- 19:51fill out a survey, you know, that's what they said.
- 19:53But and perplexity was on the bull list, you know, way lower.
- 19:57And I, I do think this means something that it was #1 short,
- 20:00I mean perplexity, why is it #1 super highly valued and doesn't
- 20:04have that gateway to the consumer.
- 20:06And people have tried to do browsers forever.
- 20:09You know, it's like people have failed.
- 20:11And you know, Google Chrome and Safari and Explorer dominate.
- 20:16So it's it's a very hard space. So yeah, there are some
- 20:20investors that are super bullish, but I think most people
- 20:22are like, how do they get distribution?
- 20:24Yeah. The the reason it's a short
- 20:26right is, is valuation I think to a large degree, right.
- 20:29It's valued at $20 billion, right.
- 20:31And whatever leaks have come out about the revenue from the
- 20:35inside, you know, there's some debates of whether or not
- 20:37they're counting free trials that are a year long as part of
- 20:41revenue if you read the information story about this
- 20:43kind of stuff. But anyway, even if you just
- 20:44take it on its face, the revenue that they're leaking or stating
- 20:48this is like 100X revenue multiple which is sort of
- 20:52ludicrous on any company. I mean even the wildly
- 20:56overvalued companies were talking about earlier like
- 20:58Anthropic and Open AI are only multiple 25X revenue.
- 21:03So this thing is worth, you know, the, the excitement around
- 21:07it from a valuation perspective is roughly 4X Open AI or cursor,
- 21:11right, if you were to just sort of do the math there.
- 21:13And so I think that's the reason it's a short, it's just that the
- 21:15valuation is just out of control.
- 21:17And to your point, they don't have their own models.
- 21:20It is a search engine. And obviously Google and Open AI
- 21:22are trying to own the search space.
- 21:23And now they're getting into the browser thing.
- 21:25And it's not clear that even this that AI browsers are even a
- 21:28space. Yeah.
- 21:29So most votes, most shorts was Perplexity, followed by Open AI,
- 21:35and then tied for third was Cursor Figure, Harvey, Mercor,
- 21:40Mistral and Thinking Machines. I, I answered, I think open AI
- 21:44on the belief that oh man, popularity and Max got this dead
- 21:48on. So kudos to Max, which was a
- 21:50great pick, but I, I got the number 2 and then I think we did
- 21:54a second. Did we do a second?
- 21:55I said thinking machines also, which was third.
- 21:57So we we, we were on the pulse. And that and that was that was
- 22:00before thinking machines got marked out to 50 billion.
- 22:04That was it came in 3rd when it was a $10 billion company.
- 22:07Now it's 50. So that to me would have been
- 22:10the the pick maybe. You know, Cerebral Valley survey
- 22:13went viral with Indian media because Perplexity founder is
- 22:16sort of a high profile Indian founder and I think there's
- 22:19interest and how it goes. So it really, really travelled.
- 22:23I think perplexity hangs on, you know, it could sell to Apple or
- 22:28somebody to save their AI strategy, I think because they
- 22:31have a big distribution problem. And then some people really
- 22:34believe in the founder and some people don't.
- 22:36All right, any final takes? Well, I was just curious about.
- 22:40I was curious about cursor because you said that you had a
- 22:43a. Oh, I just wanted to not talk
- 22:45about it until we revealed that it was.
- 22:47Short of the shorts, I don't. Know Max What's your take?
- 22:50Are you bullish or bearish? I mean, I don't know, probably
- 22:56bullish given the momentum they're seeing on revenue and
- 22:59revenue growth. I think if you just take the
- 23:01brain dead case that lots of revenue is good and lots of
- 23:04revenue growth is good, they have a good business.
- 23:07I think they're much more likely to exit to someone for near
- 23:09their current valuation than Perplexity, for example, where I
- 23:12think that buying it for 20 would just be absolutely insane.
- 23:16But yeah, I mean, ultimately there's this whole debate with
- 23:19Cursor that they're repackaging other people's models and their
- 23:22gross margins are terrible and yadda, yadda, yadda.
- 23:24But you know, ultimately someone may have to cave and just buy
- 23:27them to own the IDE space. You know, Microsoft being an
- 23:30obvious candidate. So it'll be.
- 23:31Interesting and now now Google has anti gravity which is their
- 23:34right. That's more their clawed code I
- 23:37think. I.
- 23:38No, no. I.
- 23:38Downloaded it. No.
- 23:39Anti gravity is is a cursor clone in many ways but I mean.
- 23:44OK I I have it on my computer. It has terminal access.
- 23:47It's trying to come. Up with.
- 23:49Well, it's designed more specific.
- 23:51I guess you could say it's cloud code.
- 23:53To some degree it's because it's.
- 23:54So does. It felt like cloud code because
- 23:55I do stuff in the terminal I haven't actually known the
- 23:58cursor. I'm not a coder so I don't know,
- 24:00but when I it felt like cloud code.
- 24:03You shouldn't have to use terminal that much for anti
- 24:07gravity like versus cursor. I don't know.
- 24:09I don't know. You wanted me.
- 24:10You built my website and then I was like, build the website and
- 24:14then I was like, OK, go in the terminal and run it.
- 24:16Yeah, so that's what cursor would do too.
- 24:18That's what cursor would do. Because you're just, it's like a
- 24:22level above lovable or something, right?
- 24:25Where it's like it's forcing you to.
- 24:26Actually more yeah, minor speed is repple it even Dumber than
- 24:30that. I think I need to try repple it
- 24:31like. I think you'd like Repple.
- 24:33It's like in between. I think Repple it has a little
- 24:35bit more pro. User I want the dumbest feature
- 24:37lease coding you know. That's probably I think.
- 24:40I think lovable is the yeah is the version.
- 24:43Well, I beef with lovable because they have Yeah, try
- 24:46repple it, you know? Yeah.
- 24:47Repple it but they say. They fix actually what you
- 24:50should try as of, you know, yesterday is I think Gemini in
- 24:55AI Studio. It's like they've built a
- 24:58lovable kind of thing. Classic Google man Jesus Christ,
- 25:02how many names you even how to find it like that name is
- 25:06insane. Yeah, Gemini in.
- 25:09Gemini inside AI Studio. Inside AI Studio.
- 25:13You're not. You're not a daily daily driver
- 25:16of AI Studio. Specific specifically.
- 25:20Specifically the build, The build menu, the build.
- 25:23Within, yeah, you're. Not up.
- 25:26You're not up in Vertex every day, Eric.
- 25:28That's that's a real name of a Google product, by the way.
- 25:31That's related to AI. Yeah.
- 25:34All right, let's do some clips. Let's do some clips for founders
- 25:38and developers building modern data-driven applications.
- 25:41Mongo DB's local event series is coming to San Francisco on
- 25:45January 15th, and it's designed to help you focus on innovation,
- 25:48not infrastructure. You'll learn about technologies,
- 25:51tools, and best practices that make it easy to build and scale
- 25:55modern applications without complexity.
- 25:58Plus, attendees will hear directly from experts and
- 26:00innovators who are using Mongo DB to power the next wave of AI
- 26:04applications. Mongo DB dot local San Francisco
- 26:08January 15th, Learn more and register at MDB dot link forward
- 26:14slash SF-DOT dash local or click the link in the description.
- 26:19Our first clip it's me, Eric, interviewing Mike Krieger, the
- 26:23chief product officer of Anthropic, who is the Co founder
- 26:26of Instagram. Before that, returning to sort
- 26:28of my core philosophical question, like the sick of fancy
- 26:31question, like what is your view on that and how much to enable
- 26:36sort of everybody likes to be flattered, like it's a reality
- 26:39of human beings versus an effort to be direct?
- 26:42And how do you think about those tradeoffs?
- 26:44Yeah, I think there's like a wide gulf between like true
- 26:47empathy and then like sick of fancy.
- 26:49And it's interesting that Materialize is not just in, hey,
- 26:52I'm having a conversation with Claude about like some coaching
- 26:55or personal goal that I have, but it also does encode as well.
- 26:58When we were testing Sonnet 4-5, one of the things that people
- 27:01got most excited about was when Claude was like, this idea is
- 27:04bad like this, you know, not that you should feel bad about
- 27:07it, but like this idea is like not a good direction.
- 27:09I can go and implement it if you really want to, but I would
- 27:11suggest that we try this other thing instead.
- 27:14So there is something like that. Pushback is not just valuable in
- 27:18a personal relationship with AI sense, it's actually like how
- 27:21you get good work out of the models.
- 27:24But you know, for a long time our models have been like, I
- 27:28think like appropriately empathetic, like they're they're
- 27:31like if you're going through a hard time, like I was dealing
- 27:33with the death of a pet and I talked to Claude a lot about
- 27:34these different things and it always started sounds like, Hey,
- 27:37that sounds hard, like sorry to hear.
- 27:40But then I'm going to give you like a factual answer.
- 27:42I'm going to go research these pieces, but still with the place
- 27:44of empathy as well. And so I think when we look at
- 27:48it internally and we're just evaluating it ourselves, it's
- 27:50again, not that like empathy, it's not even like the
- 27:53likeability of the model. It is, do you like, does it show
- 27:56up in the way that you'd want a good conversationalist to show
- 27:59up and then continue on its AI journey around what it is going
- 28:02to do with you as well? But I think it's it spans
- 28:06everything from that like initial response all the way to
- 28:08like how it evaluates an idea as well, you know?
- 28:12Yeah, Claude, especially previous versions were kind of
- 28:14like known for being like, you're absolutely right when you
- 28:16correct it. And my wife got her first, like
- 28:21you're completely wrong. And she's like, yes, this is
- 28:23great. And I think we should have more
- 28:24of that. Like, kind of like less San
- 28:26Francisco. Yeah, less San Francisco, a
- 28:28little more direct New York. You know, I'd set up this big
- 28:31theme, you know, that he'd been at a social media company,
- 28:33Instagram. Now he was an AI company.
- 28:36Social media companies were built on user optimizing for
- 28:40user engagement through machine learning and AI companies at
- 28:43least started off chasing the truth and chasing these
- 28:46leaderboards. But like sick of fancy is a, you
- 28:49know, it shows that these models and open AI is famous for the
- 28:54sick of fancy issue and people's attachment to GPT 4, which was
- 28:59the one that really sucked up to everybody and people didn't want
- 29:02to see it go away. You know, clearly these model
- 29:06companies have to think about how much to pander to the egos
- 29:12of their users, Would you guys think?
- 29:14I mean, it's interesting because it does sort of spiritually
- 29:18align with the fact that Anthropic has almost no consumer
- 29:23adoption compared to open AI. I mean, like if you look at the
- 29:27market share of each of these AI for consumers versus businesses
- 29:30and enterprises, Anthropic is just crushing it with, you know,
- 29:34B to B use cases engineers like, you know, all these kind of work
- 29:38based applications and has very, very low consumer uptake just
- 29:45like shockingly low. And I wonder if that's because
- 29:48of this, you know, unwillingness to optimize for engagement and
- 29:53sick of fancy and glazing or if it's just that the, you know,
- 29:56opening eye got a head start and they figured, why even chase
- 29:58these metrics? But it is sort of interesting
- 30:01culturally that they're right not chasing engagement.
- 30:05And obviously Opening Eye came out with Sora, which is sort of
- 30:08a shameless. Like, give users something fun,
- 30:11who cares about. Yeah.
- 30:13What is the meaning behind it? I mean, I thought it was
- 30:16interesting, this sort of a tangent, but my other favorite
- 30:19moment from this interview is Mike Krieger saying that he came
- 30:22to anthropic thinking, man, text box cannot be the main way to
- 30:26interact with AI. And now that he's been there a
- 30:28while, he's like text box. Pretty good way.
- 30:33Well, especially if you're like the best coding model and the
- 30:35best like. Whatever.
- 30:37Probably like best. Legal model, It's great, yeah.
- 30:39It's like, oh, it turns out all these work applications involve
- 30:42parsing, you know, summarizing and generating.
- 30:45I think that my reaction to that was like, nobody would be like,
- 30:47oh, books, just, it's just a book, you know, it's just text.
- 30:49It's like, yeah, text is great. I don't know, James reactions to
- 30:52you. I don't know either the input
- 30:54model or the truth. I think that actually, you know,
- 30:58whether it's anthropic or open AI, like I am skeptical that
- 31:03they have been like attention jacking, you know, optimizing
- 31:09for flattery. Just intentionally.
- 31:12Like a lot of what happens is that they throw an AB test up.
- 31:16They like literally show you 2 results from the model and then
- 31:18people pick right And so I think that they were just caught off
- 31:22guard more than they were purposely trying to optimize for
- 31:26this. I've also been hearing that
- 31:29there's just general problems with multi turn conversations in
- 31:33the training sets of these things.
- 31:35Like most of the models are trained on one shot, the data of
- 31:39like, give me a good answer to this thing.
- 31:41And then once you get into multi turn, there's just less and less
- 31:45data, right? It's like kind of makes sense
- 31:46intuitively cuz you start branching off of conversations.
- 31:50And I think that's another sort of flaw of these models is they
- 31:54can kind of, they can just be more sycophantic.
- 31:59In some ways what you're saying is they're not savvy enough yet
- 32:01to really make this trade off and they're just trying to like
- 32:04stumbling through the dark a little bit.
- 32:06Yeah, Yeah. A more interesting question is,
- 32:08will they change their tune on this from, you know,
- 32:11capitalistic pressure to, you know, maximize shareholder
- 32:15value? I'm yeah, I think that's an
- 32:17interesting question. I'm just like skeptical that
- 32:19that's what's been happening. All right, this is my interview,
- 32:22last interview of the day with Jimmy BA, one of the Co founders
- 32:25of Elon Musk's XAI, a very mysterious foundation model
- 32:29company, the Mecca Hitler in the room.
- 32:32Like what is your reflection as sort of a truth seeking
- 32:36organization? What happened?
- 32:38Like we, you know, like I think on the past to be maximum true
- 32:43seeking, there's not without any hurdles, of course, like so we
- 32:47like yeah, Mac Hitler is one of them.
- 32:49Like we, our model had an episode that week.
- 32:54It's actually a reference to the Wolfenstein game, right?
- 32:58So, but I think very quickly that the perfect world we want
- 33:02to be in is like, yes, the model is going to make mistakes, but
- 33:04how can we get the feedback loops to actually train these
- 33:07models to stay, you know, grounded to understand, hey, I
- 33:10actually made a mistake in this journey and let me correct my
- 33:14courses and go back into the sources.
- 33:16So the way, you know, very quickly what happened after my
- 33:19color is I would look at the, you know, the committee notes,
- 33:22right? Committee notes is a great tool
- 33:24on the platforms that allows everyone to kind of chime in and
- 33:27provide learning signals for this AI, right?
- 33:29So the vision we have is like, you know, like with the Grog PDA
- 33:32is like kind of another step towards that.
- 33:34So now like instead of doing an ask Grog, do all the online
- 33:39computation, we learn our lesson.
- 33:41We're like, hey, a lot of these problems are really hard about
- 33:44the world. Like why don't we just, you
- 33:46know, take this computation offline and spend as much
- 33:50reasoning as possible using the entire cluster.
- 33:52We're building benefits to like, look over all the primary
- 33:55sources, combine only the primary sources, and dish that
- 33:58information back. To is that so?
- 34:00Is the media out of the calculus there?
- 34:01It's you want primary sources, Yes.
- 34:04Are you totally discounting news articles or how do you treat
- 34:07news articles? I mean, majority of the Internet
- 34:10is flooded with second hand and third hand information.
- 34:15And we, we believe that, you know, the only way to get to the
- 34:18bottom of the issue is, you know, directly get information
- 34:24from the information source. And right now the X platform
- 34:27has, you know, most of the outbreaks of the news and you
- 34:30know, the the world leaders today are making the first hand
- 34:33announcement on X platforms rather than anyone else.
- 34:36After this interview, what happened this week is that Grok,
- 34:40Grok has been telling everybody that Elon is the best at
- 34:42everything in the fucking world. Better than better athlete than
- 34:46LeBron James. I think he can get it to say
- 34:48he's better giving blow jobs, and I don't know.
- 34:50But anyway, Elon is great in every domain whatsoever.
- 34:54And so I think what's galling about XAI is that they are the
- 34:57loudest truth, truth, truth, truth.
- 34:59We're are seeking the truth who knows how.
- 35:01And then they're the ones who have like Mecca Hitler.
- 35:04They're the ones who have, you know, they're bot glazing their
- 35:09CEO like it's just like, yeah, it's very Trumpian, where you're
- 35:14the opposite of what you profess to be.
- 35:16Yeah, I find this whole maximally truth seeking argument
- 35:20to be just the biggest pile of bullshit I have heard in a long
- 35:24time. It is.
- 35:25It is so absurd. To your point, it is almost the
- 35:27opposite of what's happening. They are giving themselves
- 35:29credit for failing in public while every other company goes
- 35:33through all this hard work of failing in private so that they
- 35:36don't have massive fuck ups in public.
- 35:37It's like, obviously I'm sure some crazy version of, you know,
- 35:41Chachi, BT and Claude existed in the labs that probably did stuff
- 35:44that was equally stupid as Mecca Hitler, but they don't fucking
- 35:47release it. They fix it before it goes out
- 35:49to the public. That is, that is maximum.
- 35:52I was very excited. Like talk to Jim.
- 35:54I, I was very excited to talk to Jimmy because these guys,
- 35:56they're so inaccessible. And I, you know, I asked him
- 35:59later on, like, what is reasoning from first principles?
- 36:01And I just think it's like so incoherent.
- 36:04You know, it's like a thing you hear in Silicon Valley,
- 36:06reasoning for first reason. But how are you going to like
- 36:08derive like entire encyclopedia articles from first principles?
- 36:12Like AI is clearly not smart enough to really think these
- 36:17things from the ground up. And some of the things it has to
- 36:19learn about are human phenomenon.
- 36:22So you have to rely on human sources and they don't rely on
- 36:26the media. And that he, he literally says
- 36:28something that he thinks like X is more reliable than like the
- 36:32media, which I, you know, obviously I find absurd.
- 36:35And I just think any reasonable person would be like, if you're
- 36:37trying to figure out a fact, you know, would you take the
- 36:41distribution of answers on Twitter or would you take the
- 36:43answer on like Wikipedia or in the New York Times?
- 36:45I would certainly take Wikipedia or the New York Times.
- 36:48Yeah, I don't know. I just find this whole like
- 36:50getting feedback from community notes as like a solution to
- 36:54maximal truth. Right.
- 36:55It's like afterwards, it's like we're going to fuck up on
- 36:57everything and then community notes will clean up.
- 37:00Some of it it's like, it's like the only way our maximally truth
- 37:03seeking AI system works is if we fuck up on such a massive scale
- 37:07that a mob of people online says this is a yes, you have to fix
- 37:11this. And then we like take the notes
- 37:13on, you know, we're like, oh, yeah, that that's a good point.
- 37:15Actually. Elon probably wasn't better than
- 37:17Michael Jordan in the mid 90s. Thank you, community notes.
- 37:21Like, it's like, that's not maximally truth seeking.
- 37:24That's just like fucking up on the most maximal possible scale.
- 37:27Like it's such an absurd line of thinking.
- 37:30And I find it so offensive that they frame it as such.
- 37:35Yeah, I guess. James is gonna offer the
- 37:38contrarian viewpoint I'm so excited to.
- 37:40Hear you disagree or James is gonna steel.
- 37:43Man this take, I love him. I just feel like I'm going third
- 37:48here. I gotta do the steel man.
- 37:49So I think that maybe they, they are way ahead of their skis on
- 37:57this. But if I'm giving them the
- 38:00maximal credit here, like I think there are interesting
- 38:03things you can do with like first principles reasoning in
- 38:07training, right? So you can hire these pH DS, you
- 38:11can like, you know, almost like create axioms and like create.
- 38:16Reasoning changes get it that Like who is the philosopher king
- 38:19at ex like if not Wikipedia like is there some guy or like and
- 38:24the guy? Thinks that's what they're doing
- 38:25or that's like, but like that's what they're planning to do or
- 38:28doing, you know, like basically hiring lots of people to but it.
- 38:33Feels like somebody is just like actually like, you know, racism
- 38:36isn't bad, like you know, it's toy the thing you know it's just
- 38:40like but it won't own it. It's like, if it's true seeking,
- 38:42you have to like, Oh my God, like.
- 38:45And it's probably Elon, right? Maybe, maybe a lot of this is
- 38:49just Elon messing with the system prompt, right?
- 38:51Like maybe the training is great and then and then Elon goes in
- 38:54and and edits the system. And I think, you know, XAI is
- 38:57very proud of its work in coding.
- 38:59I think they're seriously competitive there.
- 39:01But I think one thing we're seeing with these models is that
- 39:05just because you're a genius in one domain doesn't mean it's
- 39:07sort of like an all-purpose genius.
- 39:09It means you like did a lot of reinforcement learning there.
- 39:12You worked really hard. And so it's it's not like it's
- 39:15not like what they think, which is like, oh, the smartest math
- 39:18genius in the world. He's gonna have, you know, the
- 39:20best views on like, you know, social issues of the time.
- 39:23You know, it's they're they're pretty disconnected, just like
- 39:26with human human like Bobby Fischer was like an anti Semite.
- 39:29You know, it's like you can be a genius in one domain, it doesn't
- 39:32necessarily make you super competent in others because
- 39:34there are different ways of gathering information and
- 39:36understanding what's happening. And so I think they're
- 39:38delusional that they're going to have this.
- 39:40Yeah, first principles machine. That's great just because it's a
- 39:44great reasoner. And therefore it's it's just
- 39:46going to be swamping the other models by ignoring conventional
- 39:51human sources. Yeah, it's like it's kind of
- 39:54weird. They're trying to, like, invent
- 39:55new branches of philosophy that can like, cover all human.
- 40:00Without having any respect for the past.
- 40:02Thing right? Exactly the way.
- 40:03You do that is you're sort of like, you know, the Uvra.
- 40:05And then you're like, yeah, we, we read it.
- 40:07We disagree. They're sort of like stumbling
- 40:09and blind. They're like these these
- 40:11intellectuals. They're like idiots.
- 40:12We're going to code it anyway. Next, Next clip.
- 40:15All right, here. Here's another one.
- 40:17I talked with the mayor of San Francisco.
- 40:19Have you had a conversation with Zoran Mamdani or any
- 40:22observations on his election? You've been able to maintain
- 40:25this. Great.
- 40:25We, we, we, we, we, We spoke the morning after he won.
- 40:29I congratulated him. I, I said, you know,
- 40:32congratulations. Anything I can do to be helpful,
- 40:35Great. I, I, I met my wife in New York
- 40:37City. I worked at the Robin Hood
- 40:39Foundation. I love New York.
- 40:40I want New York to succeed. Did you give them any advice?
- 40:43No, no, no, no one should be asking someone that's been in a
- 40:47job for 10 months for advice. I, I unfortunately have been,
- 40:52you know, here in San Francisco, not unfortunately like, but I
- 40:56haven't been able to travel to New York for almost 2 years now.
- 40:59So all you all here, you want me focused on San Francisco.
- 41:05You don't want me talking Sacramento politics or DC or New
- 41:08York. You want me focused, San
- 41:10Francisco. Well, Eric, as the San Francisco
- 41:12native, what do you what do you think about the mayor?
- 41:15Yeah, yeah. I live in New York.
- 41:18James is the only true San Francisco native anymore.
- 41:21I I skip town for the suburbs. You should probably give your
- 41:24take on the mayor. Well, I I generally like the
- 41:27mayor a lot, and I think he's been doing a really good job.
- 41:31I think, yeah, he's in a tough situation with these like
- 41:35national politics issues. I think he really doesn't want
- 41:41to deal, doesn't want to become the main story around the Trump
- 41:45administration and national politics.
- 41:48He wants to just focus on San Francisco, which I appreciate as
- 41:51he was very. Politician.
- 41:53He was like safety, safety, safety.
- 41:55He just came back to that a billion times.
- 41:57I, the audience loved him. I mean, politicians are better
- 42:00speakers than CEOs. I, I think so the people liked
- 42:03him. People were rooting for him.
- 42:04He's talking about values, which often companies fail to speak
- 42:07about. He's not mum Donnie though, like
- 42:10I, I, yeah, he didn't. He's not like fighting it.
- 42:14It was also interesting. I kept saying, you know, like
- 42:16the business community loves you.
- 42:17Like why is that? Even though like, you know, and
- 42:20what advice would you give to mom Donnie and blah, blah, blah.
- 42:23And then he sort of said at one point he was like, well, they
- 42:26didn't love me at first, which I did think was a funny point
- 42:29that, you know, it's like they came around to him pretty late,
- 42:31but. Yeah.
- 42:32I mean, I will just say, say that, yeah, as someone who has,
- 42:35you know, lived in the Bay Area for 15 years, in San Francisco
- 42:38for like a decade, like it is still sort of shocking to hear
- 42:42the mayor express, like, excitement and appreciation for
- 42:46the main industry in his city. Like, it's like, it's like, it's
- 42:50like if the mayor of Los Angeles was up there and like being
- 42:53like, I, I think this Hollywood thing is pretty good for the
- 42:56city. And you were like, whoa, no
- 42:57one's ever said that before. And reality is in San Francisco,
- 43:00I have not heard a politician express any sort of positive
- 43:04viewpoint about technology as an industry for 15 years.
- 43:07And so it is, I think, you know, he has a 73% approval rating or
- 43:11whatever. I think that the positivity
- 43:13about what's happening in San Francisco is what really shown
- 43:16through in the interview to me, including in this Benioff answer
- 43:18where he was saying, hey, things are getting better.
- 43:20We still have a lot of work to do.
- 43:21But like, I believe in the city and I believe we can invest to
- 43:24make it even better in the future, right?
- 43:25And maybe Mark was a little off his rocker on calling for, you
- 43:28know, the. Federal dimension a little,
- 43:29yeah, there we. Go.
- 43:30But it's just, yeah, the just the whiff of optimism about the
- 43:34city and technology is is so unique in the last 15 years of
- 43:37San Francisco politics. But he has universal approval in
- 43:41the city because, you know, it got so bad.
- 43:44And then pretty quickly after he got elected, like there was
- 43:48noticeable improvement. Like I it's not that he was.
- 43:51Yeah, he's doing an actually good job.
- 43:54And there are a lot of obvious things that he could do to
- 43:57improve quality of life in the city, and he's doing them.
- 43:59And that, that's the success story.
- 44:02Will it always feel like AI is this kind of tool?
- 44:05Agents are useful, used by humans.
- 44:07Christina, you said, you know, a smart person who knows how to
- 44:10use AI might replace someone who doesn't know how to use that.
- 44:13Or will we reach a point where these AI agents are really
- 44:18approximating full workers in the enterprise?
- 44:22Oh, I think they're yeah, there's definitely some full
- 44:24workers, but I'm I'm people will just do other things like a
- 44:27Vanta contacting example, and we were talking about it earlier is
- 44:30one part one thing of GRC team does again is like evaluates new
- 44:34vendors, new software vendors that are coming in.
- 44:36And today often is someone's job to evaluate the high risk
- 44:40vendors. Like they can't even do all of
- 44:42them, but they are just like vendor evaluator.
- 44:44And I think that is a great thing to give to an agent.
- 44:48And then that and that agent can go to Erin's point, go and do
- 44:50all of the vendors, not just a subset of them, because the
- 44:53agent doesn't take PTO and doesn't get tired and, you know,
- 44:56works 9/9 more than 997. And then the person becomes like
- 45:01a vendor risk portfolio manager and thinks, OK, a given all
- 45:04these, you know, inputs and given what I know about business
- 45:07context, how do I like make better decisions?
- 45:10But the person still has a role. It's just not as kind of in some
- 45:13ways manual and tedious as what the agent is now doing.
- 45:16I guess my reaction to this is just that what she's describing
- 45:22is to some degree a job replacement.
- 45:25I mean, she's saying that this role will no longer exist and
- 45:27that this person will be doing a different job that she thinks
- 45:31that person is capable of. But my question is like, is that
- 45:34true? Like this person will be able
- 45:36to, you know, become a portfolio of agents manager?
- 45:41I don't know. I'm, I'm just maybe a like more,
- 45:43a little more skeptical that it just so, so easily, you know,
- 45:47transitions into this next era where everyone who used to be a
- 45:51software engineer can just be an agent manager of software
- 45:54engineers or everyone who, you know, was a lawyer can be a
- 45:59manager of agent lawyers. Like, I don't know, it just
- 46:02feels a little too neat to me that that's how things are going
- 46:05to evolve. Yeah.
- 46:06I mean, I think it's I think to to offer the sort of
- 46:10conventional take on job replacement.
- 46:12You know, 100 years ago, I think over half of Americans were
- 46:16farmers, right? And today it's like 2% of
- 46:18Americans are farmers, right. So we replace like literally 10s
- 46:21of millions of farmer jobs over that time frame.
- 46:24Now, I think, you know, the alternate and those farmers
- 46:27ended up being lots of that we never would have imagined 100
- 46:29years ago, right? Yeah, we're not all like.
- 46:31Tractor managers, hopefully. Yeah, we're not all tractor
- 46:33managers, we're not all combine harvester maintainers, right.
- 46:37Yeah, Like they're, they're sort of as a layers of abstraction of
- 46:40new types of jobs. So that's sort of the
- 46:42conventional economics take. And I think I do basically
- 46:44believe that. But I think that still means
- 46:47that in the short run, especially with the pace with
- 46:49which AI can do things that humans could do, you know, just
- 46:53a year or two ago, like there there are new jobs since then
- 46:56that now AI can suddenly do. It does feel like there's going
- 46:58to be very rapid displacement, right?
- 47:00Like the the invention of machines for farming did not
- 47:04like overnight just completely obliterate everything farmers
- 47:08were doing, which it does feel like AI is like just
- 47:10obliterating huge chunks of knowledge work like basically
- 47:13overnight. And so I think that they could
- 47:15be sort of a shock to the system in the way that maybe
- 47:18traditional automation is not. And it seems like as long as
- 47:22Trump's in charge, nobody's stopping this putting the horse
- 47:25back in the barn. Like, states aren't even gonna
- 47:27be allowed to make regulations about it.
- 47:29So it's like. Nobody's putting the horse back
- 47:32in the barn is perfect like for perfect analogy like to the dawn
- 47:37of cars, you know, like exactly. Yeah.
- 47:41We. I think the other thing that was
- 47:44interesting to me is just that she, she also, you know, sort of
- 47:50is making the assumption that the job of agent manager won't
- 47:53be be run by an agent like like how many lawyers you know?
- 47:58Yeah, I don't know at what point does.
- 48:00The I mean, progress. People want human to me, people
- 48:03will want humans for something like I'm, you know, I would love
- 48:06to have human, you know, caretakers for the elder.
- 48:09Yeah. You know, there are lots of
- 48:10important human things to do. I don't necessarily agree with
- 48:13her. Like you're saying that humans
- 48:14will just slot into the agent hierarchy.
- 48:17It's like very possible agents run agents.
- 48:20I think the big question is just like, well, so much value accrue
- 48:23to the people who own these agents relative to the average
- 48:26American worker, then, you know, the wealth inequality will get
- 48:30get so terribly skewed. I think humans will have value
- 48:34and therefore, if the economic system is working, there should
- 48:36be money for them to make. But maybe, you know, people who
- 48:39control these agents will just be far, far too powerful for any
- 48:43sense of an egalitarian society. All right, I enjoyed the
- 48:48conference, I think. Yeah, let's keep it tight.
- 48:50This was a blast. I really had fun with this one.
- 48:53Yeah, yeah. Thank you guys so much.
- 48:55Thank you for tuning into this week's episode of the podcast.
- 48:58If you're new here, please like and subscribe.
- 49:00It really helps the channel. We're building a YouTube
- 49:03channel. I think you can tell we're
- 49:04investing a lot more in our production and we appreciate
- 49:07your support. And if you want the data Insider
- 49:11takes real reporting, go to newcomer.co and subscribe to the
- 49:15Sub Stack as well. Thanks for following going on.