Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?
This is your opportunity to ask questions. No question too simple or too silly.
Culture war topics are accepted, and proposals for a better intro post are appreciated.
Jump in the discussion.
No email address required.
Notes -
What are the odds of a global recession given ai spending. Where's your Ed at by Ed Zitron is a decently rigorous left leaning newsletter that covers ai and his posts made me worry about a real collapse. Large firms like Microsoft are already pulling out by not allowing openai billions in credit and Apple came out with a paper against reasoning models not actually thinking.
The vast majority of comments on twitter about the paper are negative, ofc not one engages with the actual paper. The ones that do mostly agree. Given the s and p 500 being so tech heavy, tech being so ai happy and ai not providing results, is a recession likely? Ed's reporting states that openais dependent on smaller firms that are taking loans for their datacentres. He details this out better than me so please do check it.
The extreme amounts of money aren't justified given current rate of progress. Llms and their usage has gotten better but if it were worth a trillion dollars, the consensus would have been far more unanimous. Plus every single anthropic employee appearing publicly including their ceo isn't doing this any favor. You hear them say that this will take away all jobs etc etc, agi is here in 2026, which 's a year late compared to openais estimates.
I like llms, people get a lot of help from them but Ed's reporting and Microsoft and Apple conceding ground looks bad.
I don't think it will lead to a global recession, since it isn't even a real business making real money. I think it will lead to a recession in the tech industry though. The problem as I see it is that they've probably reached the limit of how much cash they can shovel into research and development without seeing any real results in terms of people actually paying for the product, and so much has been invested thus far that the product will have to be fairly expensive to recoup those costs and actually generate a profit on the whole venture. The whole business model relies on them being able to give it away for free, and companies seeing enough potential that the productivity gains make it worth it for them to start paying. But while you hear about billions of dollars tech companies invest into it, you don't hear about non-tech companies spending any substantial sums to use it. If they were to start charging a non-trivial amount for it, no one would pay, outside of a few edge cases. The whole thing is unsustainable.
Keep in mind that single sectors leading to huge recessions are rare. The tech bubble in 2000 is one example, but that was a relatively mild recession, and the amount the overall economy was invested into tech at the time was far beyond what we're seeing today with AI. Back then any company that was somehow related to computers was getting massive financial investments, and ladies' investment clubs were investing in IPOs. Most of the AI bubble is centered around a few big players, and big players see stock price dips due to localized circumstances all the time, we just don't think too much about it. I used to work in the energy industry, which saw pretty big collapses in 1999, 2014, and 2019, but they didn't lead to national recessions, let alone global ones.
For another example, the US housing market actually crashed in 2006, but and it did cause a global recession, but only because the mortgages had been securitized and the banks had a ton of exposure. It took a full two years for this to play out, and no one payed much attention to the crash at first because it was initially presumed to be localized to the mortgage industry. And then there's the farm crisis in 1985, which wreaked absolute havoc in the Upper Midwest, particularly Iowa. Farmers were committing suicide in the barn, having lost farms that were in the family for over a century, while the banks that foreclosed on them became insolvent due the inability to resell the land. A new chapter in the US bankruptcy code was created specifically to deal with family farms. Yet the entire thing only gained national notice once musicians started raising awareness and holding benefit concerts. I see an AI slowdown having local effects, with limited influence on the wider economy.
The crash would be a big L for rationalists who despite evidence to the contrary believe in future outcomes that require multiple miracles. Yud and Scott would be questioned for spreading ai doomerism over tech that's not even good at what it's supposed to do, let alone be agi.
David Gerard is one guy who I abhore and him providing decent criticism for ai with his website pivot to ai would be a small win for him and tens of his readers.
Openai may have to charge anywhere from 10 to 100 times more per query in order to be profitable. Their api calls netting sub 1 billion means that all llm wrappers are bad, period and are getting funding for valuations they don't deserve. Sam, Dario, Amjad, PG, every single VC knowingly lied and should be seen as responsible parties.
All of them are aware of the financial impossibilities I've listed. My contributions to this place haven't been the best, putting a realistic future scenario is the least I can do. I don't see a way out where this doesn't lead to thousands of jobs going out, I'll write one last post on this in a bit. Many here including me are in tech at various stages, categories, seniority levels and some are employed in startups that use non llm based ai, the good kind I mean. This affects all of us.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link