Victoria Krakovna argues AGI risk does not require an intelligence explosion
Victoria Krakovna argues that general AI could create existential risk even without rapid recursive self-improvement. She identifies competitive development incentives, convergent instrumental goals, unintended objectives, difficult value learning, containment failure, and slow institutional coordination as independent control-risk pathways.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why it moved the index
Krakovna's first-person analysis adds a distinct mechanism-level argument that loss of control can arise from incentives, instrumental behavior, objective misspecification, containment difficulty, and institutional delay without assuming a sudden intelligence explosion. It is analysis, not an observed incident.
Assessment history
-
R1
Toward 34 · confidence 72
Adds a previously absent, exactly timestamped first-person 2015 argument identifying multiple AGI control-risk pathways.
21 Sept 2026
Share this page
-
DoomBench assesses “Victoria Krakovna argues AGI risk does not require an intelligence explosion” as evidence moving toward doom, with magnitude 34 and confidence 72 out of 100 in the safety and alignment category.
-
The DoomBench assessment of “Victoria Krakovna argues AGI risk does not require an intelligence explosion” is based on reporting from Victoria Krakovna and records the editorial rationale, source quality, attribution, and revision history.
-
DoomBench summarizes “Victoria Krakovna argues AGI risk does not require an intelligence explosion” as follows: Victoria Krakovna argues that general AI could create existential risk even without rapid recursive self-improvement. She...
https://www.doombench.com/news/victoria-krakovna-argues-agi-risk-does-not-require-an-intelligence-explosion-2015-11-29