Hyperagents
cs.AI
Submitted: 2026-03-19
Updated: 2026-11-01
Comments: Code at https://github.com/facebookresearch/Hyperagents
Code: https://github.com/facebookresearch/Hyperagents
License: http://creativecommons.org/licenses/by/4.0/
The gist: Self-improving AI systems aim to reduce reliance on human engineering by learning to improve their own learning and problem-solving processes.
Terminology
Abstract
Self-improving AI systems aim to reduce reliance on human engineering by learning to improve their own learning and problem-solving processes. Existing approaches to self-improvement rely on fixed, handcrafted meta-level mechanisms, fundamentally limiting how fast such systems can improve. The Darwin Gödel Machine (DGM) demonstrates open-ended self-improvement in coding by repeatedly generating and evaluating self-modified variants. Because both evaluation and self-modification are coding tasks, gains in coding ability can translate into gains in self-improvement ability. However, this alignment does not generally hold beyond coding domains. We introduce hyperagents, self-referential agents that integrate a task agent (which solves the target task) and a meta agent (which modifies itself and the task agent) into a single editable program. Crucially, the meta-level modification procedure is itself editable, enabling metacognitive self-modification, improving not only the task-solving behavior, but also the mechanism that generates future improvements. We instantiate this framework by extending DGM to create DGM-Hyperagents (DGM-H), eliminating the assumption of domain-specific alignment between task performance and self-modification skill to potentially support self-accelerating progress on any computable task. Across diverse domains, the DGM-H improves performance over time and outperforms baselines without self-improvement or open-ended exploration, as well as prior self-improving systems. Furthermore, the DGM-H improves the process by which it generates new agents (e.g., persistent memory, performance tracking), and these meta-level improvements transfer across domains and accumulate across runs. DGM-Hyperagents offer a glimpse of open-ended AI systems that do not merely search for better solutions, but continually improve their search for how to improve.
Sources
- What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversity
- Learning to Continually Learn
- SIMA 2: A Generalist Embodied Agent for Virtual Worlds
- Quality-Diversity through AI Feedback
- Retrieval-Augmented Agentic Rubric Generation for Reliable Medical Response Evaluation
- AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence
- TICKing All the Boxes: Generated Checklists Improve LLM Evaluation and Generation
- RelevAI-Reviewer: A Benchmark on AI Reviewers for Survey Paper Relevance
- Foundation Model Self-Play: Open-Ended Strategy Innovation via Foundation Models
- Go-Explore: a New Approach for Hard-Exploration Problems
- Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution
- From Tabula Rasa to Emergent Abilities: Discovering Robot Skills via Real-World Unsupervised Quality-Diversity
- Teaching Large Language Models to Reason with Reinforcement Learning
- LLM-First Search: Self-Guided Exploration of the Solution Space
- Open-Endedness is Essential for Artificial Superhuman Intelligence
- Discovering Temporally-Aware Reinforcement Learning Algorithms
- Eliminating Meta Optimization Through Self-Referential Meta Learning
- Motif: Intrinsic Motivation from Artificial Intelligence Feedback
- Measuring AI Ability to Complete Long Software Tasks
- The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents