Is the AI Bubble Deflating?

The search for “the next new thing” has been rather unproductive over the past decade or so:

• Blockchain was a nifty technical achievement looking for a meaningful problem to solve, as was cryptocurrency.
  • The two met, and made a new style of equity for sale, which has been a bit of a boondoggle, and not nearly as transformative or crime-free as its initial promoters hoped
• Web3 was a nothingburger of the fast food variety, coming and going in nearly record time — so much so that most of us barely remember its time in the meme-0-verse.
• Machine learning has proven to be useful in a number of ways, but in the same way that better programming is always useful — it helps developers solve problems.
• Various devices — from smartwatches to VR/AR headsets — have been more style than substance overall, mixed bags even when they work as advertised, and easily eschewed when their novelty wears off.
  • Meanwhile, the unsung hero — AirPods and their offspring — go barely noticed.

And so we come to AI, GAI, LLMs, and all that jazz.

In a recent report entitled, “Gen AI: Too Much Spend, Too Little Benefit?” from Goldman Sachs, the main upside they see is either speculative or achieved via a bubble that they feel will remain pumped full of hot air long enough for some to make money. But the overall tone is that of someone coming to terms with an unpleasant reality:

Tech giants and beyond are set to spend over $1tn on AI capex in coming years, with so far little to show for it. So, will this large spend ever pay off? MIT’s Daron Acemoglu and GS’ Jim Covello are skeptical, with Acemoglu seeing only limited US economic upside from AI over the next decade and Covello arguing that the technology isn’t designed to solve the complex problems that would justify the costs, which may not decline as many expect.

In the report, Acemoglu — an MIT economist — is particularly unimpressed, arguing that “the upside to US productivity and, consequently, GDP growth from generative AI will likely prove much more limited than many forecasters expect,” and that “many of the tasks that humans currently perform . . . are multi-faceted and require real-world interaction, which AI won’t be able to materially improve anytime soon.”