jon down - where ai is going and why it poses a risk to humanity
PDF · 26 Nov 2025
This slide deck by Jon Down presents a cautionary overview of Artificial General Intelligence (AGI) and the path toward Artificial Superintelligence (ASI). It argues AGI is different from past technologies because it can learn, self-improve, make decisions and is effectively a black box; this creates both enormous potential (abundance, health) and significant existential risks. Slides include an "intelligence curve" visual, an AGI landscape, a prediction that superintelligent AGI is likely within 25 years, a list of failure modes under "Aligned Goals" (e.g., source grab, value drift, runaway growth, weaponization), and a typology of public figures grouped by their assessed risk stance.
Questions this document answers
Likely within 25 years (or sooner).