The Real Study
Rather than a generic "MIT and Stanford" joint study across 50 industries, the most credible current data point is Stanford HAI's 2026 AI Index Report, alongside Stanford research highlighted by MIT Technology Review's "reality check" coverage in May 2026. The actual findings are more specific — and more concerning for one group in particular — than vague utopian-or-dystopian narratives suggest.
Who's Actually Affected
The clearest, most consistent finding is about early-career workers in AI-exposed roles. Employment among software developers aged 22–25 has fallen nearly 20% since 2024, even as employment among their more senior colleagues has continued to grow. More broadly, Stanford's analysis found a 13% relative decline in employment for early-career workers in the most AI-exposed jobs since generative AI tools became widely adopted. The pattern repeats in customer service, another highly AI-exposed early-career field.
What Hasn't Happened
Despite the sharp, concentrated effect on entry-level tech and customer-service roles, Stanford's researchers found scant evidence that AI has yet produced a large-scale impact on the broader US labor market. Mid-career and senior workers in the same exposed fields have, on the whole, held steady or grown.
What Employers Expect Next
Forward-looking signals suggest the effect may widen: in employer surveys cited in the same research, about one-third of respondents said they expect workforce reductions over the coming year.
The Honest Takeaway
AI's labor impact in 2026 is real but narrow so far — concentrated heavily on entry-level workers in the most exposed fields, rather than a uniform "robots taking all the jobs" story or, conversely, a "nothing is happening" story.
Sources
- A reality check on the AI jobs hysteria — MIT Technology Review
- 2026 AI Index Report: Economy — Stanford HAI
- Inside the AI Index: 12 Takeaways from the 2026 Report — Stanford HAI
What the Stanford Data Actually Shows
The 2026 Stanford AI Index (published April 2026) draws on multiple datasets — BLS employment data, LinkedIn economic graph, McKinsey Global Institute surveys, and academic labour economics research — to present the most comprehensive annual picture of AI's labour market impact. Key findings:
Complementarity dominates so far. In sectors with high AI adoption (finance, technology, professional services), employment has not contracted — it has grown, with wages rising for workers who demonstrate AI tool proficiency. The productivity gains from AI have, so far, largely expanded output rather than directly replaced workers in these sectors.
Substitution effects are concentrated. Job losses attributable to AI tools are most concentrated in specific task categories: routine document processing (paralegal review, data entry, certain accounting functions), first-tier customer service, and some categories of software testing. These are tasks that can be fully specified in a workflow — the AI equivalent of what ATMs did to bank teller transaction processing.
Education level matters differently than expected. Prior predictions assumed AI would primarily displace lower-skilled workers. The 2026 data shows significant AI impact on certain high-education tasks (legal research, medical image analysis, software code completion) while many lower-education physical roles (plumbing, electrical work, general construction) remain largely unaffected by current AI tools.
The Economist Consensus
Among academic labour economists, the current consensus (reflected in a 2025 survey of the IGM Forum) is that AI is accelerating task automation in specific cognitive domains but that the labour market adjustment — reallocation to new roles, expansion of service sectors serving newly productive workers — is occurring as it has in previous technology transitions. The net employment effect is assessed as likely positive to neutral over a 10-year horizon, with concentrated negative effects in specific occupational categories in the near term.
Commenting is currently unavailable on this article.