Still don’t see how it can both decimate white collar work, which would mean huge productivity gains for the companies who adopt it, yet not be profitable? I’m in the Gary Marcus camp (ditto Freddie De Boer) who argue it’s just not good enough to replace humans (hallucinations, errors in reasoning, etc.),
AI is destined to become, more than anything, a big government subsidized pork barrel like the military-industrial complex. A recession will help; maybe the Lehman Bros of AI companies can be nationalized. A lot of the AI industry will always be fake, but that’s no problem – our economy has been largely fake ever since our most ambitious minds discovered it’s easier to innovate in hyperreality than in reality. So AI it will blend into the background eventually, like nuclear energy did. Most good AI tools predate the current bubble anyway (e.g. Google Translate) and have been AI-washed. Similarly the few benefits we get from our bloated US Department of AI will be things other companies and government orgs used to do. Just history repeating.
It will make software engineering dramatically faster and less expensive. If you are in business of building or buying software, you have learned this in the last few weeks. That’s sufficient for a business model.
The data center boom feels a bit like the fracking boom more than a decade ago, with an equal dose of the .com bubble; a heavy dose of speculation folded into a social and environmental disaster.
The comparison to deindustrialization is especially sharp: lots of diffuse upside, but the concentrated downside lands on specific workers and communities who are then told to “adapt” with almost no real support.
I think the binary you sketch at the end, either mass unemployment if AI “succeeds,” or a painful bubble burst if it doesn’t, captures why so many people feel uneasy even as they’re told this is the next electricity!
find everything you said to be really interesting but to keep things “in the realm of the dollar”, AI is already gross margin profitable and every single company and financial statement shows that demand currently outstrips supply. FT is even reporting how AI has already increased TFP. this may not continue for good with regards to future capex, but the business model is already there
What a great, rollicking read. Thank you.
Still don’t see how it can both decimate white collar work, which would mean huge productivity gains for the companies who adopt it, yet not be profitable? I’m in the Gary Marcus camp (ditto Freddie De Boer) who argue it’s just not good enough to replace humans (hallucinations, errors in reasoning, etc.),
I tend to agree. But I am already seeing it eat whire collar work so I try not to downplay the damage.
Though do we have any jobs data? Unemployment rate remains low.
I hope the government has enough money to bail out AI so soon after it bails out crypto. Unless it’s the other way around.
AI is destined to become, more than anything, a big government subsidized pork barrel like the military-industrial complex. A recession will help; maybe the Lehman Bros of AI companies can be nationalized. A lot of the AI industry will always be fake, but that’s no problem – our economy has been largely fake ever since our most ambitious minds discovered it’s easier to innovate in hyperreality than in reality. So AI it will blend into the background eventually, like nuclear energy did. Most good AI tools predate the current bubble anyway (e.g. Google Translate) and have been AI-washed. Similarly the few benefits we get from our bloated US Department of AI will be things other companies and government orgs used to do. Just history repeating.
It will make software engineering dramatically faster and less expensive. If you are in business of building or buying software, you have learned this in the last few weeks. That’s sufficient for a business model.
The data center boom feels a bit like the fracking boom more than a decade ago, with an equal dose of the .com bubble; a heavy dose of speculation folded into a social and environmental disaster.
The comparison to deindustrialization is especially sharp: lots of diffuse upside, but the concentrated downside lands on specific workers and communities who are then told to “adapt” with almost no real support.
I think the binary you sketch at the end, either mass unemployment if AI “succeeds,” or a painful bubble burst if it doesn’t, captures why so many people feel uneasy even as they’re told this is the next electricity!
Good job!
find everything you said to be really interesting but to keep things “in the realm of the dollar”, AI is already gross margin profitable and every single company and financial statement shows that demand currently outstrips supply. FT is even reporting how AI has already increased TFP. this may not continue for good with regards to future capex, but the business model is already there