The narrative that AI is wholesale eliminating jobs is overstated, according to a CNN Business report published May 10, 2026. The piece, citing McKinsey research and a string of corporate restructurings, argues that AI is mostly automating slices of work — not whole occupations — even as headline layoff counts climb.
The data behind the new framing
McKinsey estimates that roughly 57% of work-related activities are technically automatable with current AI technology, but very few jobs map cleanly onto a single automatable task. "It's very few jobs that are actually entirely automated away," McKinsey's Alexis Krivkovich told CNN. Incedo cofounder Nitin Seth offered a memorable line on the limits of arithmetic-style automation: "You can't take one quarter of Lisa, one quarter of Jessica...make it one person."
The technology is, however, deeply embedded in how people already work. A September Google research survey cited in the report found 90% of tech workers use AI in their jobs, while Stack Overflow's developer data showed 84% of respondents either use AI tools in the software development process or plan to.
The corporate signals are mixed
The same article tallies more than 49,000 job cuts attributed to AI so far in 2026. Block — the parent of Square and Cash App — cut roughly 40% of staff, Coinbase announced a 14% workforce reduction, and Cloudflare reportedly saw a 600% increase in internal AI usage over three months, with executives citing efficiency gains as a factor in headcount decisions.
Software engineering is the most visible case study. The piece argues coding is only one slice of an engineer's role, which also covers system design, code review, troubleshooting, and product decisions. One expert quoted in the article predicted the term "software engineer" may eventually give way to something like "builder" as writing lines of code becomes a smaller part of the job.
Why the distinction matters
For workers, the practical implication is that fluency with AI tools is becoming a baseline expectation, not a differentiator — and that resilience comes from owning the non-automatable parts of a role: judgment, taste, system thinking, and customer context. For employers, the picture is harder. If AI compresses many roles by 20–40% of the work, the math pushes toward leaner teams even without dramatic mass layoffs, which is roughly the pattern Block, Coinbase, and Cloudflare are signaling.
The CNN piece doesn't claim AI is benign for labor markets. It argues the failure mode is subtler than mass replacement: a slow reshuffling of what each job actually is, with restructurings layered on top — closer to a creeping reorganization of work than a single jobs-pocalypse moment.
