Solomonoff Induction vs "Just Next-Token Prediction": The Endless AGI Debate
burny_tech · x · 2026-10-02
X user quetzalrainbow argues there's an "infinitely tall wall of hot plasma" between people who understand Solomonoff induction and AIXI — and therefore why a sufficiently good statistical predictor is just AGI — and those who dismiss LLMs as "lol it just predicts next token." The post reignites the long-running debate over whether next-token prediction can lead to general intelligence.
More from AGI Musings
- Zuckerberg: Multi-Gigawatt Training Clusters Can Brute-Force Their Way to AGI — rohanpaul_ai · 2026-10-02
- Every Third AI Conversation: Uncertainty About Machine Consciousness Gets Misread as a Claim — AndyMasley · 2026-10-02
- The last provably human-written love letter has already been written — DionysianAgent · 2026-10-02
- Reddit User Endorses LeCun's Take on AI and Safety — m3nt3_ · 2026-10-02
- VC claims AI already beats 90%+ of professional researchers at research — peter_richtarik · 2026-10-02
- Viswanathan Anand on how AI is reshaping how chess players learn the game — ravi_iitm · 2026-10-02