Why?
Well, the ability for AI to replace humans seems to have been greatly over-exaggerated, with Acemoglu stating that many of these tasks “are multi-faceted and require real-world interaction, which AI won’t be able to materially improve anytime soon”.
Key to understanding this is that large language models, like ChatGPT, are not the same as artificial general intelligence, and likely are not capable of reaching the kind of reasoning and understanding needed for it to do more than generate text back at you like they currently do.
As companies increasingly spend more and more on processing power, Acemoglu questioned “what does it mean to double AI’s capabilities?” as it is unclear if that will actually be able to make it better at certain tasks. While the AI might be quicker, will it be two times ‘better’?
Along with processing capabilities, power usage is perhaps the key limiting factor of AI, which so far companies like Google and Microsoft are struggling to address.
Jim Covello, head of global equity research at Goldman Sachs, stated that the total cost being pumped into AI over the next several years, such as in data centres and utilities, will cost a trillion dollars.
He then asks: “What trillion dollar problem will AI solve?”
While some have argued that technology often starts off as expensive and should get cheaper over time, Covello described this as “revisionist history”, and said “the tech world is too complacent in the assumption that AI costs will decline substantially over time.”
He compares the technology to the internet, stating that “even in its infancy, the internet was a low-cost technology solution that enabled e-commerce to replace costly incumbent solutions”.
“AI technology is exceptionally expensive, and to justify those costs, the technology must be able to solve complex problems, which it isn’t designed to do,” he added.
In the meantime, massive tech companies will only “garner incremental revenue” from AI, Covello said, with the more time passive without significant productivity gains, the more challenging “the AI story will become”.
“Investor enthusiasm may begin to fade”, Covello concluded, if “important use cases don’t start to become more apparent in the next 12-18 months.”