PublicLedger
Gary Marcus: With focus on world models, reliable reasoning, and explicit knowledge, AGI may be achievable in the next decade (by ~2036). — PublicLedger