A social path to human-like artificial intelligence
arXiv:2405.15815 · doi:10.1038/s42256-023-00754-x
Abstract
Traditionally, cognitive and computer scientists have viewed intelligence solipsistically, as a property of unitary agents devoid of social context. Given the success of contemporary learning algorithms, we argue that the bottleneck in artificial intelligence (AI) progress is shifting from data assimilation to novel data generation. We bring together evidence showing that natural intelligence emerges at multiple scales in networks of interacting agents via collective living, social relationships and major evolutionary transitions, which contribute to novel data generation through mechanisms such as population pressures, arms races, Machiavellian selection, social learning and cumulative culture. Many breakthroughs in AI exploit some of these processes, from multi-agent structures enabling algorithms to master complex games like Capture-The-Flag and StarCraft II, to strategic communication in Diplomacy and the shaping of AI data streams by other AIs. Moving beyond a solipsistic view of agency to integrate these mechanisms suggests a path to human-like compounding innovation through ongoing novel data generation.
17 pages, 2 figures, 1 box
References in corpus (17)
- Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- Flamingo: a Visual Language Model for Few-Shot Learning
- Human-level performance in first-person multiplayer games with population-based deep reinforcement learning
- Training Compute-Optimal Large Language Models
- Exploration in Deep Reinforcement Learning: A Survey
- An Algorithmic Perspective on Imitation Learning
- The Curse of Recursion: Training on Generated Data Makes Models Forget
- A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning
- Improving alignment of dialogue agents via targeted human judgements
- Intrinsic Motivation and Automatic Curricula via Asymmetric Self-Play
- Will we run out of data? Limits of LLM scaling based on human-generated data
- Open-Ended Learning Leads to Generally Capable Agents
- Scalable Evaluation of Multi-Agent Reinforcement Learning with Melting Pot
- Mind's Eye: Grounded Language Model Reasoning through Simulation
- Asymmetric self-play for automatic goal discovery in robotic manipulation
- Quantifying the effects of environment and population diversity in multi-agent reinforcement learning
- Shaking the foundations: delusions in sequence models for interaction and control