When will the first general AI system be devised, tested, and publicly announced?

Current estimate
Dec 2032
Forecast Timeline

Top Key Factors

• Delays: Reliable >50-step agent chains with published evaluations (101)
• Hastens: AI companies don’t get funding due to AI bubble bursting (153)
• Delays: AI demonstrates ability to apply knowledge from training to novel/unseen scenarios (40)
• Hastens: China starts a war with Taiwan (90)
• Hastens: Strong regulation of AI companies by the US government (6322)
• Delays: Genuinely good US government policy on AI safety (52)
• Increases Uncertainty: The Act of forecasting changes the forecast itself (0)

Summary of Comments

Prediction Insights

• skmmcj: The upper bound has been narrowed to January 1, 2080 due to community predictions being below 2065 for most of the question's lifetime.
• t.maturkar: Forecasts adjust based on reactions; definitions of AGI remain contentious.
• tpgood1990: DARPA aims to design training for AI experts, aligning with definitions of AGI from industry leaders.
• AussieA: Sees AGI as dependent on hardware scaling issues, predicting late 2030.
• Davidlwarren: Discusses key indicators of generalization and adaptability in AI.
• alextes: Shares that breakthrough definitions of AGI may remain elusive.

Notable Predictions

• Jgalt: Predicts 2029 (2027-2030), pointing to Boston Dynamics' Atlas humanoid robot partnership with Google DeepMind.
• BlueDot8338: Expounds on the role of AI in automating software engineering, predicting breakthroughs in 2027.
• Manda_Ralijaona: Forecasting early 2032, highlighting physical limitations on power grids as bottlenecks.

Conclusion

Current discussions highlight the essential debate around timelines and definitions tied to the progress of general AI, emphasizing factors like funding, technological advancements, and regulatory environments.