AI Post Transformers • Visual Companion

From AGI to ASI and Beyond

This page treats DeepMind's report as a regime map rather than a proof: AGI is drawn as a threshold, ASI as an organization-scale claim, and the gap between them as a mix of compounding pathways and missing measurements.

arXiv:2606.12683 Genewein et al. • DeepMind • Posted June 10, 2026 AGI as stepping stone, not endpoint Transcript IDs: scanning...
Tab 01

Regime Map

The paper's core move is structural: human-level generality becomes an intermediate state, while four reinforcing pathways point toward organization-scale capability.

Lens
Tab 02

Concept Grid

Some ideas are mathematically sharp but operationally distant; others drive forecasts and policy conversations despite softer formal grounding.

View
Tab 03

Compounding Compute

These curves are mock forecasts shaped by the transcript's multiplier logic: hardware, capital, and algorithmic efficiency compound quickly, while organization-level usefulness lags under friction.

Scenario
Line Emphasis
Tab 04

Evidence Gap

Current benchmarks mostly measure task skill and partial autonomy. The rightmost claims in the ASI story remain the sparsest columns.

Comparison Class
References

Cited Papers with arXiv Links