AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence

Abstract

Perhaps the most ambitious scientific quest in human history is the creation of general artificial intelligence, which roughly means AI that is as smart or smarter than humans. The dominant approach in the machine learning community is to attempt to discover each of the pieces required for intelligence, with the implicit assumption that some future group will complete the Herculean task of figuring out how to combine all of those pieces into a complex thinking machine. I call this the "manual AI approach". This paper describes another exciting path that ultimately may be more successful at producing general AI. It is based on the clear trend in machine learning that hand-designed solutions eventually are replaced by more effective, learned solutions. The idea is to create an AI-generating algorithm (AI-GA), which automatically learns how to produce general AI. Three Pillars are essential for the approach: (1) meta-learning architectures, (2) meta-learning the learning algorithms themselves, and (3) generating effective learning environments. I argue that either approach could produce general AI first, and both are scientifically worthwhile irrespective of which is the fastest path. Because both are promising, yet the ML community is currently committed to the manual approach, I argue that our community should increase its research investment in the AI-GA approach. To encourage such research, I describe promising work in each of the Three Pillars. I also discuss AI-GA-specific safety and ethical considerations. Because it it may be the fastest path to general AI and because it is inherently scientifically interesting to understand the conditions in which a simple algorithm can produce general AI (as happened on Earth where Darwinian evolution produced human intelligence), I argue that the pursuit of AI-GAs should be considered a new grand challenge of computer science research.

References in corpus (53)

• Deep Learning in Neural Networks: An Overview
• Sequence to Sequence Learning with Neural Networks
• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
• Overcoming catastrophic forgetting in neural networks
• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Practical Bayesian Optimization of Machine Learning Algorithms
• Neural Architecture Search with Reinforcement Learning
• How transferable are features in deep neural networks?
• Trust Region Policy Optimization
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Understanding Neural Networks Through Deep Visualization
• On the Properties of Neural Machine Translation: Encoder-Decoder Approaches
• Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
• Continual Learning Through Synaptic Intelligence
• Evolution Strategies as a Scalable Alternative to Reinforcement Learning
• Human-level performance in first-person multiplayer games with population-based deep reinforcement learning
• Evolution of Biological Complexity
• Noisy Networks for Exploration
• Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning
• RL${}^2$: Fast Reinforcement Learning via Slow Reinforcement Learning
• Illuminating search spaces by mapping elites
• Learning to reinforcement learn
• Parameter Space Noise for Exploration
• Large-Scale Study of Curiosity-Driven Learning
• FeUdal Networks for Hierarchical Reinforcement Learning
• Go-Explore: a New Approach for Hard-Exploration Problems
• Episodic Curiosity through Reachability
• Building high-level features using large scale unsupervised learning
• Visualizing and Understanding Neural Models in NLP
• Visualizing and Understanding Atari Agents
• Learning to Optimize
• Third-Person Imitation Learning
• Emergent Complexity via Multi-Agent Competition
• Convergent Learning: Do different neural networks learn the same representations?
• Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions
• Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks
• The evolutionary origins of hierarchy
• Illuminating Generalization in Deep Reinforcement Learning through Procedural Level Generation
• Meta-Learning and Universality: Deep Representations and Gradient Descent can Approximate any Learning Algorithm
• Understanding Deep Image Representations by Inverting Them
• Learning Montezuma's Revenge from a Single Demonstration
• Learning to Continually Learn
• Imitating Latent Policies from Observation
• A Theory of Universal Artificial Intelligence based on Algorithmic Complexity
• Unsupervised Meta-Learning for Reinforcement Learning
• From Nodes to Networks: Evolving Recurrent Neural Networks
• Diffusion-based neuromodulation can eliminate catastrophic forgetting in simple neural networks
• Backpropamine: training self-modifying neural networks with differentiable neuromodulated plasticity
• The Emergence of Canalization and Evolvability in an Open-Ended, Interactive Evolutionary System
• Resolving the paradox of evolvability with learning theory: How evolution learns to improve evolvability on rugged fitness landscapes
• Building Machines that Learn and Think for Themselves: Commentary on Lake et al., Behavioral and Brain Sciences, 2017

Cited by in corpus (30)

• Solving Rubik's Cube with a Robot Hand
• Unity: A General Platform for Intelligent Agents
• Multi-task learning for natural language processing in the 2020s: where are we going?
• Learning to Continually Learn
• Discovering Reinforcement Learning Algorithms
• Improving Generalization in Meta Reinforcement Learning using Learned Objectives
• Open-Ended Learning Leads to Generally Capable Agents
• Generative Teaching Networks: Accelerating Neural Architecture Search by Learning to Generate Synthetic Training Data
• Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutions
• A Comprehensive Overview and Survey of Recent Advances in Meta-Learning
• The whole brain architecture approach: Accelerating the development of artificial general intelligence by referring to the brain
• BADGER: Learning to (Learn [Learning Algorithms] through Multi-Agent Communication)
• Meta-Learning with Sparse Experience Replay for Lifelong Language Learning
• Qualities, challenges and future of genetic algorithms: a literature review
• Dynamics-Aware Quality-Diversity for Efficient Learning of Skill Repertoires
• Scaling Down Deep Learning with MNIST-1D
• Finding online neural update rules by learning to remember
• Grounding Artificial Intelligence in the Origins of Human Behavior
• Fiber: A Platform for Efficient Development and Distributed Training for Reinforcement Learning and Population-Based Methods
• Self-Organizing Intelligent Matter: A blueprint for an AI generating algorithm
• Meta-Learning of Structured Task Distributions in Humans and Machines
• MPLP: Learning a Message Passing Learning Protocol
• Semi-Supervised Class Discovery
• Training Learned Optimizers with Randomly Initialized Learned Optimizers
• A Brief Survey of Associations Between Meta-Learning and General AI
• Evolutionary Self-Replication as a Mechanism for Producing Artificial Intelligence
• User-Oriented Smart General AI System under Causal Inference
• Toward Building Science Discovery Machines
• A method of supervised learning from conflicting data with hidden contexts
• Metalearning: Sparse Variable-Structure Automata