Redditor's six-axis AGI framework scores frontier models: only reasoning passes, memory and continual learning fail
johnc1100 · reddit · 2026-11-15
A Redditor, unconvinced by current AGI claims, built a six-axis AGI framework with 0–4 levels (refined with Gemini 4 Hydrogen into a webpage) and scored the frontier:
- System 1 control (sub-50ms closed-loop perception-action, rapid physical adaptation): 2/4
- System 2 reasoning (formal puzzles + open-world causal simulation): 3/4 — the only axis at the human bar
- Autonomy (indefinitely stewarding endless tasks, preventing error drift): 2/4
- Metacognition (knowing what you don't know, mid-task pivots, real-world verification): 2/4
- Long-term memory (decades-spanning episodic memory, O(1) retrieval, selective forgetting): 1/4
- Continual learning (sample-efficient permanent updates without catastrophic forgetting): 1/4
Conclusion: by this definition frontier models remain far from AGI (level 3), with memory and on-the-job learning the weakest links.
More from AGI Musings
- Ex-Citizen Quant Head Dan Morillo: AI Outputs Consensus, Not Alpha — minhnguyen · 2026-11-15
- 104 job titles across 80 years: an interactive family tree of software roles from 'Computer' to 'Context Engineer' — bibryam · 2026-11-15
- LeCun's CoRL slide: 'If you are interested in human-level AI, don't work on LLMs' — Heavy-Ad9232 · 2026-11-15
- RL trade-offs will split LLMs into role-specific personalities, argues Redditor — uriwa · 2026-11-15
- "Double-blind review doesn't work": reviewers see author names, and rival students reject competitors — Temporary_Switch_339 · 2026-11-15
- Betsey Stevenson's post-AGI talk: a 10% income rise lifts well-being the same everywhere — MartinSignoux · 2026-11-15