When tech entrepreneur Matt Schumer posted his warning about artificial intelligence wiping out lawyers, doctors, financial analysts, and journalists, the internet did what it does best: it panicked. The post has since racked up more than 80 million views, spawning a wave of career anxiety among workers who fear they are one software update away from redundancy.
There is just one problem. The evidence for this sweeping disruption is almost entirely absent, according to Bloomberg.
@bloombergopinion A list of all the ways #AI will decimate #jobs went viral. The truth is much more boring, Parmy Olson says. #chatbots #AIagents #tech
♬ original sound - Bloomberg Opinion
Schumer's post, for all its reach, rests on anecdote rather than data. It gestures at professions under threat and urges readers to spend an hour a day practising with AI tools to stay ahead of the curve. What it does not offer is concrete proof that mass displacement is coming anytime soon.
What the numbers actually show
US productivity statistics, the clearest signal of AI transforming how work gets done, remain up slightly and comfortably within their historic range. That is not the picture you would expect if millions of knowledge workers were being replaced or radically augmented by software.
The Yale Budget Lab, which has been tracking the labour market since ChatGPT launched, has found no discernible disruption to broad employment patterns. Jobs are not disappearing at an unusual rate. New ones are appearing. The economy continues, more or less, as before.
Perhaps most revealing is a randomised controlled trial examining AI tools among experienced software developers, the workers most often cited as early casualties of the technology. The finding was counterintuitive: developers using AI assistance actually took longer to complete tasks than those working without it. The productivity miracle, at least in this controlled setting, did not materialise.
Klarna's cautionary experiment
The case of Klarna, the Swedish fintech company, offers a useful corrective to the most breathless AI predictions. Klarna made headlines when it announced it had replaced 700 customer service workers with AI. It was cited widely as proof of what was coming for everyone else.
Then Klarna started rehiring.
The details of why remain murky, but the basic fact is instructive: deploying AI at scale to replace humans turned out to be more complicated and less reliable than the initial announcement suggested. Reality, as it tends to, proved messier than the headline.
Trading on vibes
What Schumer's post captured, and what made it spread so far so fast, is not evidence of disruption. It is the anxiety that surrounds it. AI is, at this moment, largely trading on vibes: a mixture of genuine capability, persistent hype, and collective uncertainty about what the technology will eventually do.
That uncertainty is real, and it is not irrational. AI is genuinely changing some work, in some places, for some people. The honest picture is that its effects are uneven, incremental, and extraordinarily difficult to predict with any precision.
The problem is that the viral panic around posts like Schumer's tends to flatten that nuance into a single terrifying narrative: that the transformation is total, imminent, and already decided.
The data, so far, disagrees.
The gap between hype and reality
None of this means AI is irrelevant or that its long-term effects on employment will be minor. It means that the current moment is not the one the panic describes. The disruption being sold to 80 million viewers is running well ahead of what the evidence supports.
Productivity gains are modest. Labour market data shows no signal. A controlled trial found AI tools can make skilled workers slower. A high-profile replacement experiment was quietly reversed.
That is not the story of a technology already reshaping the economy. It is the story of a technology with significant potential that has not yet delivered it at scale, surrounded by an enormous amount of noise from people who have every incentive, financial or psychological, to convince you otherwise.
The best advice for workers worried about AI is probably not to spend an hour a day using chatbots. It is to watch the data rather than the discourse, and to treat any claim that rests entirely on vibes with the scepticism it deserves.