The total spending on generative “AI” since 2013 is something like $2.5 trillion. In perspective, this is like taking all of the economic activity in a country like Spain or Canada over the course of an entire year and redirecting it entirely into the manufacture of chips and data centers and the programming and training of LLMs. An incredible amount of resources.
7
Replies
·
0
Boosts
@HeavenlyPossum The interstate highway system is an interesting comparison point because it suggests that, for a similar investment, the USA could have built high-speed rail parallel with the interstates instead of ‘AI’. That’s something that economists will happily tell you is predicted by models to have a huge positive economic impact across the nation.
What does the world get in return for that massive investment? Worse than nothing:
“In our in-progress research, we discovered that AI tools didn’t reduce work, they consistently intensified it. In an eight-month study of how generative AI changed work habits at a U.S.-based technology company with about 200 employees, we found that employees worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day, often without being asked to do so.”
“…this report uncovers a surprising result in that 95% of organizations are getting zero return.”
https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf
In contrast, the US spent a mere $298 billion on its Apollo space mission and got something tangible in return: giant rockets that carried human beings to another world and back.
Why is the world getting *so little* from its investment in generative AI, especially compared to other investments that cost less but produced far more?
Setting aside the fact that AI is largely a scam and that the finances of AI are a titanic bubble waiting to pop at any moment, I think that discrepancy in scale is worth considering. Where else have we seen something like that?
https://www.reuters.com/graphics/USA-ECONOMY/AI-INVESTMENT/gkvlqbgxkpb/
Consider that the Holland Tunnel opened in 1927, connecting New Jersey and Manhattan. It cost about $48 million to construct, equivalent to about $890 million today.
Consider that the Hudson Tunnel Project, an ongoing effort to build and revitalize rail tunnels to New York City, is currently estimated to cost more than $16 billion.
Why has infrastructure gotten so much more expensive to construct in places like the US?
There are a variety of ways to answer this question. Labor has gotten more expensive. Regulations have proliferated. Raw materials are harder to source. And so on.
There’s a kind of meta-answer we can also answer, one that encompasses all the rest:
Complexity.
Social complexity is a problem-solving tool. Societies often add complexity when they confront new problems: new specializations, new layers of bureaucratic management, new interconnections between components of those societies.
But every additional layer of complexity requires resources. A general doctor has to be fed and clothed by the labor of other people while they are performing the labor of medical care.
Expand medical care from general medicine to a whole ecosystem of highly specific specialists, though, and you now have many more people to feed and clothe while they are performing their specialized care.
Critically, the marginal returns on those specializations decline. The very first doctor in a community might not be able to treat every ailment, or treat them very effectively, but a generalist provides a very high marginal return on those resources invested. These are the low-hanging fruit of complexity.
But a hyper-specialized doctor might only be able to provide care to a small number of people, in highly specific settings. That care is very important to the patients receiving that care, but the marginal returns on those invested resources are lower.
