AI companies, AI influencers and famous professors have been making extraordinary claims for years about AI. It would put radiologists, doctors, and lawyers out of work, cure cancer, make nuclear fusion and other new science-based technologies economical, enable one-person companies, create thousands of new billionaires, and dramatically improve productivity and per capita income.

A few months ago, many even claimed that humanity had advanced more in the previous 3 weeks than the previous 100 years because of OpenClaw, Opus 4.6, Codex 5.3, and MiniMax. Readers probably don’t remember those predictions but five months ago they were touted by the AI tech bros as evidence that AI was moving at warp speed. One post making this claim on X.com received almost 900,000 views

It wasn’t just the tech bros and AI influences who have been impressed by these extraordinary claims, but also investors. These claims have been inflating the bubble for years. The bubble was purported to be $35 trillion last October; since then, the Nasdaq has risen by 20% and the number of trillion-dollar companies has now reached 10. Unfortunately, the claims have almost always been made without much evidence, despite that assertion by astrophysicist Carl Sagan and others that extraordinary claims require extraordinary evidence.

One reason that investors, CEOs, government ministers and other elites are misunderstanding the impact of AI is because they’re not close enough to the work to understand how well AI does. According to the CEO of Box, CEOs are “sufficiently distant from the last mile of work that still has to happen to generate most value with AI.” In other words, “AI-happy CEOs are out of touch with the rank-and-file workers tasked with making their AI ambitions come to life.”

Surveys have demonstrated these gaps in perceiving the impact of AI. One survey found a gap between executives and non-management staffers. A January Wall Street Journal article reported “two-thirds of nonmanagement staffers said they saved less than two hours a week or no time at all with AI. More than 40% of executives, in contrast, said the technology saved them more than eight hours of work a week.”

Surveys have also found this to be true between CEOs and board members, the latter being even further from real work. “61% of CEOs say their boards are rushing AI transformation, exposing a divide at the top just as companies enter a critical phase of scaling AI. The CEO’s go further: “75% of board members believe their AI knowledge is on par with or ahead of that of their peers, CEOs are less convinced. Nearly 40% say boards lack an informed view of how AI is reshaping growth strategy, and one-third say boards overestimate the human capabilities that AI can replace.”

Government ministers are likely underestimating the challenges of implementing AI even more than CEOs and board members. The Trump administration has repeatedly made superficial claims about AI, and in the UK, the newly appointed chief secretary to the Treasury, Lucy Rigby, said that not using AI in public services would mean ‘choosing decline’,

One way that CEOs, executives, government ministers, and other elites underestimate the challenges of implementation of AI is that AI and other forms of automation involve a lot more human involvement than most realize.

For instance, self-driving cars require humans to wait in the background to take control of the cars when there is a problem. Rodney Bookes, roboticist and Roomba inventor, has long said that people are always behind AI, functioning as a “human in the loop” to smooth out errors, fix edge cases, and make these systems truly practical. Many CEOs ignore this work, however, and aren’t willing to look below the surface.

Unfortunately, those kinds of details are largely unknown because “many executives and managers operate at an abstract level, working via spreadsheets, emails and Zoom meetings.” They often know little of “concrete labor, meaning the specific, friction-heavy tasks that workers perform, like writing code or wiring server racks. When a board-room full of execs loses sight of this tangible labor, by failing to consider the kinds of tasks AI chatbots are actually good at, it can certainly create a break from material reality, though one driven by social factors rather than psychological.”

Some of this hype may have extended to fraud, which I talked about in my article last month for Mind Matters. Even the Wall Street Journal suggested there might be fraud in an article entitled: “Can Investors Trust AI Sales Figures.”

It shouldn’t be surprising that hype and fraud would be increasing right before OpenAI and Anthropic do their IPOs, particularly because of their big losses. Some claim that OpenAI loses $1.69 for every Dollar of revenue and that Anthropic has similar losses. Furthermore, these losses have caused AI software companies to raise their prices in the hope that they will look better to investors when they do their IPOs, which has exacerbated the cost of tokens for AI users.

The hype may be backfiring now though as companies pull back from excessive usage, according to a Wall Street Journal article. “Just a few months ago, the prevailing sentiment around AI use at many big companies was the more, the better.” But those “all-you-can-eat subscriptions amounted to a subsidy by the model-makers, which often lost money on the intensive activity of power users. Exhorted to embrace the wave of change, employees at some companies engaged in tokenmaxxing, or using as much computing as possible in order to be seen as AI-forward—a practice that continued even as the model companies shifted to usage-based pricing.”

Even Meta, a provider of cloud computing services, is involved in the pullback. “It has been great to let people experiment but now we have too many overlapping tools,” Meta’s CTO Andrew Bosworth wrote in a memo to employees. “Nobody should be using AI tools just for the sake of using them. All motion is not progress and token usage alone is not a measure of impact of any kind.”

Will the bubble pop or is the pullback another false alarm? In any case, the extraordinary claims will likely continue and perhaps even accelerate as AI startups push their IPOs and companies become fearful of the bubble popping.