The AI hype cycle is pulling capital away from the problems that matter, and the cost will arrive long before the technology does.
EXPERT OPINION BY TOM CHI, FOUNDING PARTNER, AT ONE VENTURES
What if, instead of new tech disrupting one industry, there was a piece of tech so advanced that it disrupted all industries? What if we could make an AI that achieved human-level intelligence (AGI), or one that was superintelligent (ASI)? Couldn’t that replace all humans and be worth the total of the global economy?
Honestly, it’s a great pitch for anyone who loves to dream about their money becoming more money, which is why that huge sucking sound you hear is the sound of nearly all of venture capital going into AI. In 2025, AI took up 61 percent of all venture dollars, and in Q1 of 2026 that share jumped to over 80 percent.
But even if the promises of AGI were true, the pitch itself raises some pretty obvious questions: What exactly is the value of “wealth” in a world that has immiserated most people and the entire planet? Are humans only worth their economic output? Are our lives anything more than labor playing out the whims of a money-maximization algorithm?
While I find the answers to these questions obvious, I’m not here to argue them today. I’d like to suggest something bolder: that we are not going to achieve AGI or ASI any time soon, but we may absolutely succeed in immiserating ourselves well ahead of technological possibility, based on our misguided allegiance to the pitch. What I see, instead of proximity to AGI, is a hype cycle looking to justify its own large-scale accumulation and mismanagement of capital.
The daily digest for entrepreneurs and business leaders
An Inc.com Featured Presentation
Stolen valor
I see this most clearly in “stolen valor”-style narratives, where people point to Waymo’s self-driving cars as evidence that their LLMs, and AI overall, are going to take over every industry. As someone who worked on that program and who is also familiar with the architecture of LLMs, I can tell you there is a big difference between these two tech stacks.
One is doing motion planning based on high-accuracy sensor fusion; the other is playing around with words to predict the next token. One was built with safety as a foundational design constraint, and is provably safer than human drivers. In July, the Insurance Institute for Highway Safety (IIHS, an insurance-industry-funded safety group) found that Waymo’s driverless vehicles deployed in San Francisco, Phoenix, Los Angeles and Austin were involved in 68 percent fewer crashes than human drivers per mile
LLMs on the other hand hallucinate, which is to say they state falsehoods with confidence, and help kids cheat on their homework. They may also be at least partially behind the slew of outages and infrastructure failures of the past week according to recent reporting by the Financial Times.
