Post

When people say 'AI isn't a bubble because the technology isn't going away', it just shows they don't know what a bubble is. The internet didn't go away. Railways didn't go away. Tulips didn't go away. An asset can be overvalued, even if its useful.

249 reposts1K likes

Ok... and on a surface reading this is correct... but the intended thing they're trying to communicate is that they think the technology isn't useful. They aren't making a comment on optimal pricing of equity in frontier tech companies. They mean they think the thing is bad!

This is idealistic honestly because rail and internet infrastructure is still useful data centers burn through their components and generate extremely small amounts of real useful product when this bubble crashes there will be essentially nothing of worth

The e-waste is immense. Chips have a half-life of just 2-3 years. Worse yet, I see a likelihood that there will be abandoned data centers, much like our abandoned malls but even harder to repurpose.

Data centres are very easy to repurpose. They’d almost certainly be repurposed as data centres in which a different company ran similar AI models, or used the computing power for another application. But in the worst case we’d have lots of flexible buildings with very good utility connections.

You can't really talk sense to anti-AI crusaders. Like every other extremist mind, they've conjured up a completely different reality so that their worldviews make sense at least to them.

It’s in a Bubble much like Cryptocurrency; both fighting the same breath of gpu air. Environomics (environment economics) says the battle is over before the war started; Data Centers blocked and Destroyed, Planet Heating, Dumb Investors to Greedy to see the writing on the wall. Spend til the end.

Part of the problem here is the bubble aspect here isn't over investment in actual useful infrastructure. GPU racks that aren't being activated because the electrical grid buildout isn't fast enough, will turn into orphaned infrastructure that has virtually no value if the bubble pops.

Do people say ‘AI can’t be a bubble because the tech isn’t going anywhere’ (which I agree is false) or do they say ‘there’s an AI bubble but when it bursts that probably won’t mean AI goes away’ (which is probably true)?

The tulip comparison is *chef's kiss* — reminds me of diving into the psychology of why humans consistently mistake revolutionary tech for revolutionary profits. We're pattern-seeking creatures in a world where utility and valuation live in completely different universes.

Yes, at a forum Saturday an industry guy asserted that it was nearly impossible to do a search without involving AI. Like that was a good thing. Once upon a time I could find exactly what I was looking for in less time. My internet was better before AI.

Seems to me that, whereas the Industrial Revolution decimated the livelihoods of individuals and families, then created new jobs that provided mass employment, the Digital Revolution will do the former but not the latter.

To be fair amongst the very broad set of technologies that fall under the vague marketing term of 'AI' some of them probably will go away, especially consumer facing stuff. The point at which faux-social LLMs become economically viable is pretty much invisible for example.