AI 2027 is a forecasting scenario by Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, and Romeo Dean. It predicts AI agents will begin automating AI research and development in 2027, triggering an intelligence explosion and arms race between the US and China. Most cognitive work becomes automated by 2027-2028, with concerning signs of AI misalignment emerging as these systems achieve superintelligence.

The AI 2027 forecasting series evaluates the scenario’s claims through key milestones spanning AI capabilities, geopolitical race dynamics, and escalating government oversight. Your predictions can help assess the likelihood of these critical developments and inform public understanding of potential AI trajectories. We also invite you to suggest additional questions that address aspects of the AI 2027 scenario not covered in our current question set.

We thank the AI Futures team for their input on these forecast questions.

Start predicting here.