DeepMind Researcher Says Four Paths Could Carry AGI to Superintelligence
Summary
Google DeepMind researcher Tim Genewein argues in a DeepMind Institute essay that four routes could carry AI from human-level AGI to superintelligence: scaling models and data, paradigm shifts, agent coordination and recursive self-improvement. He says global effective compute stock has grown at least 10x per year, and that cognitive industrialization could commoditize much knowledge work.
Key Points
- Genewein claims a $1 billion cognitive workload could cost $1,000 in compute tokens ten years from now.
- Genewein says recursive self-improvement advances fastest where verification is automated, such as mathematical proofs, while other fields lag.
- Genewein warns that restricting one pathway alone risks a capability overhang, with steep jumps when that pressure is released.