Aeon — An AI-native programming language
summary
Aeon is an AI-native language that synthesizes code by combining LLMs + SMT + genetic programming and then verifies results with liquid types (SMT-backed refinements) so generated functions must satisfy concrete constraints like “HTTP status ∈ [100,599].” Examples: a DSL plus genetic programming explains individual neural net predictions for genomics, Super Mario levels are generated with hard and soft constraints, and an AutoML/datapipeline verifier catches shuffling/imbalance and data leakage. Claims include 84 real-world bugs modeled and finding several drone/robotics bugs in an evening.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 5:05AI perceptions and developer challenges
- 13:13Neural networks limitations in bioinformatics
- 18:13SMT solvers in verification
- 22:52Combining AI techniques in Aeon
- 31:25AI system bugs in robotics
- 36:49Q&A
related talks
24mNatural Language Coding Revolutionising Programming with GenAI
32mBuilding an AI Operating System
33mAutoGen: A Multi-Agent AI Framework
26mThe present and future of programming with AI (Keynote)
05mAI is your children, what is our future with them?
39mA2A in 45 Minutes: Building Interoperable AI Agents
42mReasonable AI
34mAI Agents: Transforming Autonomous Decision-Making