Over a year ago, Daniel Kokotajlo (former OpenAI), Scott Alexander and colleagues published “AI 2027” – a detailed, month-by-month scenario exploring how AI progress, alignment challenges and geopolitical dynamics could unfold.
It presents two plausible paths (a “Racing” and a “Slowdown” Ending) and has since been updated with a systematic review of its 2025 predictions plus an interactive timelines model.
For anyone seeking a concrete, evidence-based foundation to think seriously about AI’s future, this is a valuable resource.
Links: • Full scenario:
https://ai-2027.com/?twclid=2c2ivesqq1162p575ofgzukie7 • PDF: https://ai-2027.com/ai-2027.pdf?twclid=2c2ivesqq1162p575ofgzukie7 • Grading of 2025 predictions (Feb 2026): https://blog.aifutures.org/p/grading-ai-2027s-2025-predictions?twclid=2c2ivesqq1162p575ofgzukie7 • Current interactive AI Futures Model: https://aifuturesmodel.com/?twclid=2c2ivesqq1162p575ofgzukie7
A factual examination of these scenarios helps us understand developments more clearly and engage with them responsibly