Richard Waters and David Rotman Debate AI’s Impact on Jobs and Productivity

In a dialogue published by MIT Technology Review on December 1, 2025, Financial Times columnist and former West Coast editor Richard Waters talks with MIT Technology Review editor at large David Rotman. Part of “The State of AI” series—a collaboration between the FT and MITTR—they examine generative AI’s effects on the job market and introduce the idea of an “economic singularity.”

Richard Waters on Uneven Adoption and the Productivity Lag

Waters notes generative AI’s rollout has been spotty. Software developers see big changes from AI coding tools; Mark Zuckerberg predicted half of Meta’s code would come from AI within a year. Yet most firms get no gains from early investments—a MIT study Waters cites found 95% of gen AI projects deliver zero return.

Skeptics point to AI’s probabilistic flaws and hallucinations as barriers to real business change. Waters counters with history: Erik Brynjolfsson described IT’s “productivity paradox” in the 1990s, where tech shifted work but aggregate productivity stayed flat until US growth jumped above 2% in the mid-1990s, only to stall later.

For AI to pay off, companies must build data platforms, redesign processes, and retrain staff. A Fortune 500 executive Waters quotes reviewed analytics use and saw workers adding little value—swapping them for AI could help, but it demands years of overhaul. Recent US productivity has climbed back over 2%, though government shutdowns delay confirmation, and AI rides prior tech waves like cloud and mobile. Robotics might amplify this further.

David Rotman’s Skepticism on Broad Gains

Rotman agrees productivity is AI’s real test, beyond hype like chatbots or agents. Brynjolfsson expects a J-curve: heavy upfront investment, then sharp gains. But IT’s post-2000s slump—despite smartphones, Slack, and Uber—shows that pattern can fizzle.

MIT economist Daron Acemoglu (2024 Nobel winner) predicts modest, slow gen AI gains because it targets narrow areas irrelevant to big sectors like manufacturing. That 95% failure rate shows the problem; big tech’s internet-trained models ignore factory-floor needs, like snapping a problem photo for advice.

Rotman urges focusing on AI that boosts workers—nurses, teachers, factory hands—rather than blaming poor implementation. Recent layoffs blamed on AI often just cut costs. True productivity, per Brynjolfsson and Acemoglu, comes from new roles and worker augmentation, not job slashing.

Wrapping Up: Cautious Hope

Waters replies they’re both wary but spots upside. McKinsey estimates 60% of work (vs. Acemoglu’s 20%) falls in AI’s reach, for up to 3.4% annual economy-wide gains—plus bonuses from job enhancements. Cost cuts lead with new tech, but AI moves quick, so early days.

The article links to further reading, like FT’s Martin Wolf on AI potentially bucking his productivity doubts (with risks of job loss and “techno-feudalism”) and Rotman’s prior work on AI’s economic effects.