Employers are under enormous pressure to adopt AI and ditch employees. Investors and CEOs fantasize about slashing costs and boosting margins; every CIO is pushed to come up with an AI plan, to keep up with competitors. Dreams of AI-agent-driven revolutions are everywhere.
But leaders shouldn’t feel like they have to rush to embrace a future that isn’t here yet. There are lots of reasons for caution. Here are nine:
“Experts” have often been wildly wrong in their predictions. The Nobel laureate and AI pioneer Geoffrey Hinton said in 2016, “People should stop training radiologists now… It’s just completely obvious that within five years, deep learning is going to do better than radiologists.” But few if any radiologists have been replaced a decade later. Google cofounder Sergey Brin promised in 2012 that driverless cars would be ubiquitous by 2017. Today, 14 years after that promise (and many subsequent ones by Elon Musk), fully autonomous vehicles remain a limited experiment, available in only a small number of fair-weather cities.
Big Tech wants you to believe it has created artificial general intelligence. That doesn’t make it true. When tech CEOs warn of employment Armageddon, they might be covering their bases in case that actually happens, but then again, maybe they just want you to drive up the valuations of their companies. Take every projection they make with a grain of salt.
When it comes to impact on employment, AI giants’ numbers don’t support their claims. Anthropic’s CEO has been warning of a jobpocalypse, but Anthropic’s own recent research showed the gap between perception and reality. The company projects great potential for what AI might do in fields like finance and architecture. But what it called “observed AI coverage” (a nice phrase for what is happening in the real world) made up a comically small fraction of that theoretical reach. What they imagine AI might do and what it is actually doing are light-years apart.
Current AI is “jagged” (good at some things but not others), which means it can seldom entirely replace a human. AI can definitely help the productivity of some workers, but even on tasks that AIs are good at, models and agents often make silly mistakes, some of which are hard to detect. And tasks aren’t jobs: Even if AI can do some part of a person’s job, it doesn’t mean it can do all of that person’s job.
Current AI models still have trouble going beyond language. Some white-collar jobs involve only words, but many involve visual comprehension: interpreting images, charts, diagrams, blueprints, maps, and so on. It might seem easy to imagine AI taking over every job, especially if you think of it as some form of magic. But once you realize that current AI is a tool, with strengths and weaknesses, you start to realize that the tech is only likely to displace workers in some professions and not others (and more often will simply augment human jobs). Even in domains like customer service that might seem straightforward, results are often disappointing. The Remote Labor Index focused on jobs that could be accomplished completely over the internet, and found that less than 4.5% could actually be adequately completed by AI agents.
Most physical labor goes well beyond what current AI can do. Don’t expect AI to replace plumbers, carpenters, auto mechanics, nurses, house cleaners, forest rangers, chefs, appliance repair workers, gardeners, or many other jobs anytime soon.
Many layoffs that have been attributed to AI aren’t really about AI. This may have been the case for the recent mass layoffs at fintech Block; some saw it as an effort by CEO Jack Dorsey to regain investors’ confidence after its stock tanked. In many cases AI may be serving as a fig leaf to cover layoffs that are actually driven by financial underperformance or earlier overhiring.
Some layoffs that are attributed to AI don’t last. I call this the Klarna Effect, after buy-now, pay-later company Klarna, which proudly made massive AI layoffs only to reverse them. Many of the people laid off worked in customer service, but after 11 months Klarna decided that (at least in some cases) “real humans” were required after all.
Overall impact on productivity and return on AI investment has so far been modest. Every company is investing in AI, but so far most aren’t getting huge returns.
All this could change; probably someday it will—but most likely not until we see more radical advances in AI, which could be a decade or more away. In the meantime, the advice is simple: Don’t focus on replacing humans. Focus on how you can use AI to help the ones you’ve got.
Gary Marcus is an emeritus professor of psychology and neural science at NYU, and the author of six books, including Taming Silicon Valley.
This article appears in the April/May 2026 issue of Fortune with the headline “9 reasons not to freak out (yet) about AI.”
