Safety and alignment

Stuart Russell launches $5.5 million center for human-compatible AI

UC Berkeley launched the Center for Human-Compatible Artificial Intelligence under Stuart Russell with a $5.5 million grant and a multi-university team to develop provably beneficial AI, formal control guarantees, and systems that learn human values instead of optimizing fixed objectives.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

CURRENT ASSESSMENT · REVISION 1
AWAY FROM DOOM36confidence 94/100

Why it moved the index

The completed launch created durable, funded technical capacity for research on value learning, formal guarantees, and human control. It is meaningful away-from-doom evidence, while its magnitude is limited because it establishes research capacity rather than demonstrating that the control problem was solved.

AUDIT TRAIL

Assessment history

  1. R1
    Away 36 · confidence 94

    Adds a source-verified historical safety-capacity milestone from UC Berkeley with an exact publication timestamp.

    21 Sept 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Stuart Russell launches $5.5 million center for human-compatible AI.
  1. DoomBench assesses “Stuart Russell launches $5.5 million center for human-compatible AI” as evidence moving away from doom, with magnitude 36 and confidence 94 out of 100 in the safety and alignment category.

  2. The DoomBench assessment of “Stuart Russell launches $5.5 million center for human-compatible AI” is based on reporting from UC Berkeley and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “Stuart Russell launches $5.5 million center for human-compatible AI” as follows: UC Berkeley launched the Center for Human-Compatible Artificial Intelligence under Stuart Russell with a $5.5 million grant and a...

    https://www.doombench.com/news/stuart-russell-launches-5-5-million-center-for-human-compatible-ai-2016-08-29