Apollo economist says AI is not yet erasing jobs

On June 1, 2026, the Superhuman newsletter pointed to an analysis by Torsten Sløk, chief economist at Apollo. According to the text, Sløk says weekly employment data show zero evidence of layoffs driven by artificial intelligence. Apollo manages more than one trillion dollars in assets, so this is not a marginal observer’s view. His argument is that AI is still creating demand for work in data centres, development, infrastructure and related services. That does not mean AI will not change work. It means the public debate about immediate human replacement may be running ahead of hard data. In 2026, two things are happening inside companies at the same time. Managers are testing automation in administration, support, software development and marketing outputs. At the same time, technology companies are building data centres, hiring specialists in models, security and integrations, and buying hardware. The net impact on employment cannot be measured by one anecdote about a team being laid off. In the same issue, Superhuman also covers Microsoft Copilot Health. The service is meant to connect wearable devices, laboratory results and health records from more than 50,000 US providers. Microsoft is relying on a partnership with Harvard Health and a panel of more than 250 doctors. From a business point of view, this is an attempt to open another strongly regulated market for AI. From a social point of view, it is an area where a model error does not carry the same weight as a badly written email. The connection between work and health shows why the AI debate will be more complex than a slogan about replacing people. In one part of the economy, AI will shorten the preparation of text or code. In another, it will bring new control processes, legal review and responsibility for sensitive data. In healthcare, the important questions are who guarantees sources, how a recommendation is explained and whether the user understands the boundary between information and diagnosis. Sløk’s argument has a practical consequence for companies. Instead of broad predictions about layoffs, it makes more sense to track specific tasks. Which activities does AI speed up? Where does it improve quality? Where does it simply add another control step? Such measurement is less dramatic than a large forecast, but it is more useful for management. If a company deploys AI without data on actual use, it will not know whether it saves time or merely moves work into output checking. Terms: Weekly employment data are short-term statistics showing new claims, hiring or layoffs. Copilot Health is a Microsoft health AI service connecting personal and medical data. A regulated market is a sector where data use and services are subject to strict legal rules.

Apollo economist says AI is not yet erasing jobs

The full article is available to users registered at Hard Skills