Sequent: scale and automation for higher confidence in alignment

Short summary

Sequent, a new AI alignment nonprofit co-founded by researchers from UK AISI and Timaeus, launches with a mission to achieve higher confidence in AI safety through theory-driven research and automation before superintelligence emerges. The founding team plans to scale to 40-80 researchers within two years, combining theoretical foundations (singular learning theory, deep learning understanding) with empirical validation to move beyond reactive alignment methods used by labs.

• •UK AISI and Timaeus researchers launch Sequent nonprofit to advance AI alignment through theory and automation
• •Targets higher a priori confidence in safety before ASI development via portfolio of parallel research bets
• •Plans 40-80 FTE in 2 years across Bay Area, London, and Melbourne; leverages automation to accelerate progress

Generated with AI, which can make mistakes.