Thoughts on Recent AI Sentiment from Frontier Labs?
Some interesting stuff coming out of Anthropic and OpenAI this week. An Anthropic researcher leaving over safety concerns, another tweeting that the chance of AI driven human extinction is >10% this decade, and Dario, Altman, and Musk all saying there needs to be a slowdown.
What are people’s thoughts on all this? Is Anthropic trying to generate hype prior to the IPO, and is OpenAI using this as an excuse to delay theirs?
Is this some form of “altruism” where a bunch of tech guys want to pretend like they are saving the world? Or maybe they are just too optimistic given their proximity to the technology?
Or do they genuinely believe we are at an inflection point? Is the threat so real that the major Frontier Lab CEOs are begging for regulation, or perhaps they just want to put up some guard rails while making sure their competitors play by the same rules?
Curious to hear people’s thoughts on this.
The threat has gone from “wow I might lose my job” to “wow all of humanity might die”
Oh well, having Claude generate three sentence emails for me was fun while it lasted
They’re utterly unprofitable, AI products are still full of errors, they gobbled up all the available spare capital and investors demand even further yields.
Marketing before IPOs. They’re also pushing to have their entire business model be regulatory capture. The want the industry regulated up the butt.
Read Situational Awareness
It is all hype-based.
OpenAI and Anthropic want more regulation/slow-down now that they have established themselves as the duopolies.
They didn’t have this same sentiment when they entered the market and all of a sudden they want to put up regulations. They all complained about government regulation hindering innovation but it only applies when it benefits them.
Plus I am assuming both companies have hit a limit on their ability to acquire/retain customers so this could be an excuse for when they IPO.
This
"OpenAI and Anthropic want more regulation/slow-down now that they have established themselves as the duopolies."
^Very good point.
Yeah. They are realizing they don't have sustainable businesses and are trying to pivot in a way that doesn't tank their equity value.
I say this as someone who believes that humanity's inevitable end state with AI is to treat the Butlerian Jihad as a checklist the way the Canadians treat the Geneva Conventions when they stop saying sorry, this fearmongering is nonsense. It is just marketing.
Dario is a such a punchable guy. He just wants regulation to protect his duopoly / monopoly. Screw him, I would double down on open source just to screw him over. Cannot describe how much I want to bash his skull in, all he talks about is people losing jobs and then AI killing us all....all to generate capital for Anthropic and regulations to lock in his forever monopoly
I HIGHLY recommend you watch the last 2 episodes of the All-in podcast (thumbnails AI Kills Us All? and OPENAI HITS AGI) or go through David Sacks' X posts over the past 4-5 days as I think he has one of the most thorough explanations of what the fuck is actually going through these people's heads after having been the government's go-to AI guy at the beginning of the current admin. NONE of this is genuine. The claim from an Anthropic executive that AI has a great than 10% chance of wiping out all of humanity (X post below) is horseshit and has the potential to earn them a visit from the SEC for possible IPO quiet period violations.
This entire thing is fueled by desire for 1) money and 2) control. Their valuations are stupidly overhyped and can't be sustained. The business model they've created is the ultimate example of "growth at all costs," and has grown to the point their capex buildout represents a not small portion of the entire US GDP growth at the moment. The private markets are tapped, so the only way left to get more money is if they go public. But their revenue growth faces a massive threat from open source competitors who can release models that are 80-90% as good as theirs at less than 10% of the cost. Meanwhile Anthropic just released their financials and a college accounting student can see their claims about being gross margin profitable on tokens has been a lie from day 1. They're literally doing the WeWork community adjusted EBITDA meme with a trillion dollar valuation.
So how do you stop the open source guys from eating your lunch and putting price pressure on you? Easy, establish a duopoly through regulatory capture. Tell everyone your tech is so dangerous and so important that the government just HAS to step in and heavily regulate AI development so that only your frontier teams are allowed at the forefront while open source players are held back.
Just think for 2 seconds - in what universe would they have been allowed to get as far as they have if there was an over 10% real likelihood that what they're doing would wipe out humanity within a decade? If they think there needs to be a slowdown on AI development - they are THE FRONTIER! They quite literally control the pace of said development. Nobody says they can't slow down. They are choosing not to slow down.
If they genuinely believed that there was a threat to the existence of the human race in their labs wouldn't they I dunno, write that down somewhere? Would it be somewhere in any of their filings? Would they perhaps sit down with the government to try and work something out? Or would they only make casual posts about it on social media to get attention and drive excitement (good or bad) about their IPOs?
ALL of the "rogue agent hacked XYZ" stories within weeks of going viral were found out to be staged. The 3rd party threat risk assessments for these labs are all done by companies with clear conflicts of interest. For example, Dario proposed 3rd party evaluators for AI risk and pushed for the use of METR. The majority of METR's assets are in the Good Ventures Foundation, which is mostly made up by Anthropic $500m of stock placed by Dustin Moskovitz which in less than 2yrs is now worth over $7B. The same organizations that fund METR also fund the Tarbell Center, which is another prominent AI doomer news outlet. The guy who just did the media tour as an ex-Anthropic "whistleblower"? He was employed at the company for a few months and wasn't even a data scientist/researcher, he was some nobody that suddenly created an X account and on his first post went viral with 120m+ impressions then got brought on every major network to spout "it's gonna kill us all" with 0 explanation of how/why that would happen.
It's also worth noting both orgs have a huge amount of involvement with the "effective altruism" movement. Know who else was a prominent tech CEO that wanted to IPO a large company with money problems while actively calling for government regulation of their competitors? SBF. Funny how things change but stay the same. "We're so wonderful and altruistic you just have to trust we have your best interest at heart, so please give us total control of this industry and regulate anyone who challenges us into oblivion pretty please." Seriously fuck these people.
I'll check out those episodes - I do have to laugh, though, at the thought of reading through a filing and within the MD&A section seeing "Key risks: additional funding doesn't materialize, increased competition, total human annihilation"
Just isn't as sexy as infinite growth, universal income, and the Jetsons lifestyle.
I didn't realize they released a bunch from their summit, so it's not the last 2 it's the ones with the titles I listed (comment above updated). The Jensen Huang interview about the doomer hoax is also worth checking out.
There is some genuine hilarity about the nonchalantness required to try and pull that off. I don't think the AI frontier nerds have that kind of motion. But they'll for sure tell you about their $30T TAM.
You know the AI bullshit is at an all-time high when I could not agree more with every single word PrivateTechquity wrote.
Truly the uniting issue of our time.
Agree with everything you say, but I'd go even further on this one. Not only regulatory capture, but also gives the big AI frontier labs the ability to continue peddling bullshit and lies by saying "well, all our promises about profitability and impact would have been true, BUT the government stepped in and so we couldn't get there!"
The entire AI boom is built on people making up the most obvious bullshit, and promising vague future results that cannot be falsified. Every new model is about to end work as we know it, or hack every computer in existence, or wipe out humanity. You know what we never hear about? How much money they make (or lose). Everything is a puff piece, no one is held to account, and this is just one more step in that direction. Shout and swing your arms around wildly about some absurd, overblown danger in order to distract from the really disturbing reality that the entire American economy is now concentrated in an "industry" which lights hundreds of billions of dollars on fire every year, all to make a product which is no more useful than a calculator or typewriter. Think about the exposure that everyday people have to AI, even through basic stock indices! When all this shit eventually crashes, hundreds of millions of people are going to be looking for heads to roll, because people are going to lose a lot of their net worth, their retirement savings, etc. It'll be 2008 again, but the difference is that this time there are a very small, very visible number of people who have been spewing the bullshit and building the house of cards, and they're the ones who will end up being vilified.
This is just the opening salvo in their gambit to disclaim responsibility when that happens. The Amodeis and Altmans of the world want to be able to look back in 5 years and say "it wasn't our fault millions of households were left destitute, it was government intervention and regulation which caused the crash!"
The funny part is many of the results they predict already have been falsified. Radiologists will be completely replaced by AI > AI penetration in radiology for scan detection is at nearly 100% but the result is more radiologists now vs before. 100% of code written within 2 years will be done by AI and will replace most engineering jobs > there is measurably more demand for software engineers and more job openings now vs before AI entered the vertical. White collar jobs will be replaced by Agents > more job openings now vs before across most sectors paired with no discernable evidence ANY white collar job has experienced meaningful worker replacement by agents. It's comical they managed to get these (by comparison) low stakes predictions so drastically wrong but this tech is supposedly a threat to all of humanity unless governments intervene. Doomers gonna doom.
Many thanks... in that case, would you recommend people in Real Estate avoid pivoting into data centers sectors? It might be a huge bubble too?
I don't know. I've seen some deals that look amazing, but my lizard brain says anything that looks amazing has some risk you're not seeing. E.g. look at what's happening with Oracle's data center contracts and the force majeure they just had to issue because local government is preventing them from setting up power supply... I'm not about to call a top and crash because the market seems to keep finding a way... but I get the sense over the next year and change we're going to see some brutal shake outs for at least a few of the big players and the tourists that just see the money being thrown around and jumped in head first.
It’s gonna be funny as hell if they go public and on their earnings call an analyst asks them about margin outlook and then follows up with their latest thoughts on AI nanobots turning us into grey goo
Illo sint at praesentium velit ipsa eos molestiae. Magnam aut est occaecati reiciendis. Et quae officia animi laboriosam. Maiores explicabo aliquam et esse.
Expedita odit veritatis quae quaerat hic laudantium. Totam earum laborum aperiam illo asperiores. Qui quod ducimus vel voluptatem corrupti a. Saepe asperiores quam et accusantium et voluptatem. Vero et sunt nulla minus officia.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Cumque corporis quaerat fuga id et. Fugit error debitis eveniet dolorem blanditiis ut.
Quos molestias qui nisi quis consequuntur non doloremque. Et et nostrum rerum sit maiores temporibus. Maxime corporis dolorem accusantium sapiente alias. Nobis neque veniam vel magnam explicabo quaerat voluptates.
Omnis enim molestiae qui et atque perspiciatis. Nesciunt eligendi tenetur occaecati inventore unde necessitatibus vero. Est quis autem cupiditate consequuntur. Et magni ut rerum unde ex.