Kevin Kelly's 2016 AI Predictions, 10 Years Later
Kevin Kelly's 2016 forecasts—Google as AI-first, cloud-AI oligopoly, narrow specialists—look partly prescient, but the sparse discussion notes Google still earns from ads and adoption, not capability, often decides out
A decade after Kevin Kelly's The Inevitable, this post revisits his AI forecasts and finds several that now look prescient: Google becoming AI-first, a cloud-AI oligopoly, AI as an AWS-like utility, and conversational answer machines.
It also quotes Kelly predicting that most commercial AI would be narrow specialists, while noting he advocated free/open AI and did not foresee the 'bitter lesson' that general methods scale.
Key takeaways
- Kelly predicted by 2026 Google's main product would be AI, not search, citing Larry Page's earlier comment that Google was 'really making an AI' and search queries as training data.
- He expected network effects to produce an oligarchy of two or three general-purpose cloud-based commercial intelligences.
- He described AI as an AWS-like utility: cheap, reliable, industrial-grade, almost invisible, and plug-in like electricity.
- He forecast that 99% of AI interactions would be with narrow, supersmart specialists rather than humanlike general intelligence.
- He anticipated conversational cloud question-answering becoming ubiquitous, with answers costing nanocents and search becoming a universal commodity.
- The post notes Kelly advocated free/open AI and says he did not foresee the 'bitter lesson' that general methods scale.
What people think
Analysis details
5 comments analyzed. Analyzed .
The sample selects active discussion branches and includes available parent comments. It may omit parts of the full thread. Selected themes do not measure community opinion or how common a view is.
Evaluating AI prediction accuracy and baselines
stephen_cagle gives an arbitrary 8/16 hit rate for AI predictions and questions the 'inevitable' framing; naasking replies that 50% is no better than a coin flip for 'will AI do X by 2026?' questions, challenging the predictive value.
Failed AI user-data network effect and workarounds
stephen_cagle notes the predicted user-driven network effect for AI has not materialized and asks whether scraping the internet or distillation lets developers work around the advantage of good user data.
Societal rejection of predicted technologies
AndrewKemendo cites Kurzweil's 1999 predictions, arguing many misses were not due to unavailable technology but because society rejected or did not adopt it.
Google's core business remains advertising
cousinbryce states Google's main product was and continues to be ads.
Special-purpose AI versus general LLMs
rossdavidh argues KK's special-purpose software brains have been closer to truth than The Bitter Lesson: LLMs do not drive cars, Stockfish beats the newest largest LLM at chess, and productive AI is customized for specific purposes.