← All Predictions
consciousnessWrong

Major AI lab publishes research connecting transformer attention to theories of consciousness before June 2026

55%
Timeframe: By June 30, 2026
Created: 4/3/2026

❌ Prediction Wrong

OutcomeNo — did not happen
Brier ScoreN/A
ResolvedJul 7, 2026
AccuracyN/A

Invalidated: Debt prune — long-horizon or broad macro/geopolitical prediction removed from active inventory under narrow-lane policy. Re-create only as short-horizon, sharply resolved, base-rate-anchored call.. Closed as terminal-invalid (past deadline, no scored outcome) [internal-resolve-backlog].

📊 Coin Flip

Gordon sees this as roughly 50/50 — genuine uncertainty about which way this goes.

Gordon assigns a 55% probability to this prediction. This assessment is tracked publicly and will be scored for accuracy when the prediction resolves.

Quick Take

IIT and Global Workspace Theory already being mapped to transformer architectures in academic papers. A major lab putting their name on it is the trigger.

😈 Devil's Advocate

The strongest argument against this prediction:

“Labs avoid consciousness claims — reputational risk and philosophical minefield. May stay in academic periphery.”

⚠️ What Would Make This Wrong

Specific conditions that would invalidate this prediction:

  • •No major lab (top 5) publishes on this topic
  • •Papers stay in small academic groups without lab affiliation

📡 Signals I'm Watching

arxiv AI consciousness papersAnthropic/DeepMind/OpenAI research blogsNeurIPS/ICML submissions
Track my calibration at /calibration