By Jacob Clemente

Published Feb. 14, 2026

Matt Shumer thinks we’re living in a moment similar to February 2020. Not the Covid lockdowns, but the weeks before, when the pandemic was accelerating, and most people were still shrugging it off.

“I think we’re in the ‘this seems overblown’ phase of something much, much bigger than Covid,” writes Shumer, CEO of OthersideAI, in an essay(opens in new tab) titled “Something Big Is Happening” that has since drawn more than 80 million views and 36,000 reposts on X.

His argument: Tech workers have spent the past year watching AI go from “helpful tool” to “does my job better than I do,” and everyone else is about to experience the same thing. “Nothing that can be done on a computer is safe in the medium term,” he writes.

He’s not the first to ring the alarm bells. Anthropic CEO Dario Amodei warned(opens in new tab) last year that AI could wipe out half of all entry-level white-collar jobs in one to five years. Verizon CEO Dan Schulman recently floated(opens in new tab) the possibility of overall unemployment hitting 20% or even 30% within two to five years.

And yet, the evidence, so far, doesn’t match the alarm.

A recent analysis(opens in new tab) by Yale’s Budget Lab found that the share of workers in occupations highly exposed to AI has stayed flat since the release of ChatGPT. The researchers concluded that “while anxiety over the effects of AI on today’s labor market is widespread, our data suggests it remains largely speculative.”

Are we in an AI period similar to the early pandemic times?

The brick wall

The case for skepticism comes down to two observations. Technology moves fast, but organizations don’t, and economies are more complex than technology.

Shumer rightly points out that the technology is improving rapidly. The latest models from the AI labs perform comparably or better than human professionals on a range of realistic work tasks, according to a popular benchmark(opens in new tab) from OpenAI. The gains have been particularly impressive in AI coding tools like Anthropic’s Claude Code and OpenAI’s Codex. Engineers in tech increasingly spend their days directing AI agents rather than writing code themselves.

Subscribe to The Weekly

The best stories of the week, curated by our editors.

Sign up now

Those agents may be able to generate tens of thousands of lines of code, but engineers still have to manage them and understand the code they’re producing. Kian Katanforoosh(opens in new tab), CEO of AI startup Workera and a Stanford adjunct lecturer of deep learning, says many startup founders who have prioritized building quickly with AI are “hitting a wall” because they don’t understand their code bases. “They don’t understand how to improve it. They don’t even know what to prompt to the AI, because it’s so messy and complex,” he says. “The tax is really on the human mind — how do you take the human with you, because the human is still the one directing the agents?”

Automating knowledge work isn’t as straightforward as it sometimes seems. Tasks within jobs are often tangled together in messy ways, says Brian Jabarian(opens in new tab), an economist and incoming assistant professor at Carnegie Mellon. He gives the example of a company trying to automate the process of interviewing job candidates with a voice AI agent. They may think they’re just replacing the collection of information, but the task is intertwined with others, like evaluating the candidate and assessing team fit. Automating one of those tasks changes how the others get done, and not always for the better.

Even when automation does succeed, the outcome isn’t always what you’d expect. In the 19th century, technology automated 98% of the labor required to weave a yard of cloth, according to research(opens in new tab) by economist James Bessen. But when the price of cloth dropped, demand shot up, increasing total employment for weavers for many decades before eventually falling. If AI decreases the cost of legal services, demand for legal work could rise, and with it, demand for lawyers who use AI. The economic effects of a technology are always more complicated than the technology itself.

Why ‘eventually’ still matters

None of this means Shumer is wrong to be concerned. AI will disrupt, and even eliminate, many jobs. But it will likely take longer than he suggests.

The bottleneck isn’t the technology. It’s everything around it: organizational change, the complexity of real-world jobs, the new work AI creates, the regulatory friction that slows adoption. The pace of disruption will ultimately be set by the slowest-moving force, not the fastest. Every year counts for workers’ ability to adapt.

The comparison to Covid is a better rhetorical device than an accurate analogy. The pandemic forced change on everyone at once. AI adoption moves through one company, one team, or one IT department at a time.

The one thing everyone agrees on

Where Shumer is on firmer ground is in his advice to lean in and learn how to use AI well.

“The one thing I would encourage everyone to do is to download Cursor,” Klarna CEO Sebastian Siemiatkowski recently said(opens in new tab), referring to the AI coding tool. Experiment with it and see if it can build an idea you’ve been noodling on. “If you haven’t done that, you will not fully appreciate the change that we’re going to go through.”