2025 Drags to a Close

As we near the end of the year I’m seeing quite a few predictions for next year, many of them either unremarkable or preposterous. This morning on a lark I asked ChatGPT what its predictions were for 2026. Here are the results:

1. AI spending hits a credibility wall in the form of pushback from boards of directors and CFOs for immediate measurable results from adopting AI.
2. A visible tiering of AI users emerges among power users, occasional users, or institutional avoiders including government, regulated professions, and unions.
3. White-collar hiring freezes spread, not layoffs
4. One major AI firm retreats from “frontier” scaling away from ever-larger models toward efficiency, specialization, or verticalization.
5. Electricity becomes a binding constraint. It should be noted that will give an edge to China in the adoption of AI.
6. Courts quietly restrict AI use in legal proceedings
7. Medical AI stalls at the liability boundary
8. A backlash against “AI fluency” hiring language
9. The first serious AI-driven outsourcing reversal appears. Work previously offshored will be reshored not to human workers but to AI.
10. Public discourse shifts from “Can AI do X?” to “Who is responsible?”

Some of those are verbatim. Some are paraphrases. If you asked the same question I suspect the answer would vary considerably. After several years of regular use on my part ChatGPT has a pretty fair sampling of how to respond to me. YMMV.

I asked several follow-up questions. I may report on those in the coming year.