This article is terrible haha it’s clickbait. LLMs have yet to provide any meaningful productivity gains. The buildout will slow once big tech runs out of cash
Hmm no. That is not right, I think. The wrong qualifier there is "any". There are definitely productivity gains, I experience them every day. There are also definite quality gains if used right.
The thing that is perceived as a bubble is the circular money flow and the valuations. It doesn't matter at this point if GPT and Anthropic were to go bust and it's all a lie. We have the gains without them, open weights, localizable. The gains will stay.
I mean quantifiable productivity gains that actually show up in revenues and profits.
Not the email, coding tech bro productivity gains.
I’ll believe AI has real productivity gains when the cost of construction goes down. If construction companies can take on more work and increase margins because of AI then I will someone agree with your comment. Until then I remain skeptical
The main cost components of construction are... materials and labor? Doesn't that mean that it is basically depending on all the things where automation has the least influence? I.e.
materials: monetary policy / tariffs = politics
material transportation: oil price = dollar strength / geopolitics
labor: supply & demand = this one you will (and I think should) disqualify because the datacenter demand bubble has outsized influence.
As an outsider, I see no reason to believe that material costs or pressure on labor will come down in any other scenario than bubble pop, or maybe only a little when a whole 180 on policies like steel tariffs or wars were to happen (unlikely.) Maybe I am wrong about that, but I think that this means that construction, especially the cost side, is one of the sectors the least correlated to the type of productivity the current automation wave is bringing as all the gains will be in the margin.
The exception could be the realization of Elon's army of Optimus or equivalent from a competitor, but that is robotics way more than the type of AI the bubble is composed of (note that that is a Tesla product, not a SpaceX product.) And besides: I expect that the whole world will be bitching when construction margins rise due to labor displacement; insider bank account NgU will not be celebrated outside that circle.
I find it more likely that any non-bubble success of AI is having an inverse correlation on construction costs due to pressures created by the datacenter demand - at least in the mid-term.
Personally, I think that the constant attention seeking narratives are mostly a shitstonk pump strategy and I agree that skepticism is absolutely warranted because I do see a lot of waste in the capital allocation towards AI (i.e. it's a bubble.) But we shouldn't be skeptical intrinsically at a technology, but rather at the integrity of the sales pitches that attach magical properties to it.
The actual technology is a generic enabler; it can break moats, walls around gardens and re-distribute asymmetric advantages. It can and already does help to increase productivity in places, but not all productivity. That's the bullshit coming out of the US labs, the AGI nightmare they pitch because fear is a great sales tactic on the uninformed.
This article is terrible haha it’s clickbait. LLMs have yet to provide any meaningful productivity gains. The buildout will slow once big tech runs out of cash
Hmm no. That is not right, I think. The wrong qualifier there is "any". There are definitely productivity gains, I experience them every day. There are also definite quality gains if used right.
The thing that is perceived as a bubble is the circular money flow and the valuations. It doesn't matter at this point if GPT and Anthropic were to go bust and it's all a lie. We have the gains without them, open weights, localizable. The gains will stay.
I mean quantifiable productivity gains that actually show up in revenues and profits.
Not the email, coding tech bro productivity gains.
I’ll believe AI has real productivity gains when the cost of construction goes down. If construction companies can take on more work and increase margins because of AI then I will someone agree with your comment. Until then I remain skeptical
The main cost components of construction are... materials and labor? Doesn't that mean that it is basically depending on all the things where automation has the least influence? I.e.
As an outsider, I see no reason to believe that material costs or pressure on labor will come down in any other scenario than bubble pop, or maybe only a little when a whole 180 on policies like steel tariffs or wars were to happen (unlikely.) Maybe I am wrong about that, but I think that this means that construction, especially the cost side, is one of the sectors the least correlated to the type of productivity the current automation wave is bringing as all the gains will be in the margin.
The exception could be the realization of Elon's army of Optimus or equivalent from a competitor, but that is robotics way more than the type of AI the bubble is composed of (note that that is a Tesla product, not a SpaceX product.) And besides: I expect that the whole world will be bitching when construction margins rise due to labor displacement; insider bank account NgU will not be celebrated outside that circle.
I find it more likely that any non-bubble success of AI is having an inverse correlation on construction costs due to pressures created by the datacenter demand - at least in the mid-term.
Personally, I think that the constant attention seeking narratives are mostly a shitstonk pump strategy and I agree that skepticism is absolutely warranted because I do see a lot of waste in the capital allocation towards AI (i.e. it's a bubble.) But we shouldn't be skeptical intrinsically at a technology, but rather at the integrity of the sales pitches that attach magical properties to it.
The actual technology is a generic enabler; it can break moats, walls around gardens and re-distribute asymmetric advantages. It can and already does help to increase productivity in places, but not all productivity. That's the bullshit coming out of the US labs, the AGI nightmare they pitch because fear is a great sales tactic on the uninformed.
PS:
coding tech bro? Go wash that mouth! lol.I agree with this response!
productivity gains would be nice. But they're not necessary - a product with demand alone is enough to grow the economy in and of itself
yup