Epoch AI | MirrorCode Benchmark: What's the largest software project AI can complete on its own?

​How large a software project can an AI system complete autonomously—not merely fix a bug or implement a feature, but rebuild an entire program?

​Join ⁨⁨Tom Adamczewski⁩ and Epoch AI for a look at MirrorCode, a benchmark co-developed with METR for long-horizon coding. In each task, an AI must reimplement a real program end-to-end without access to its source code, then match its behavior on held-out tests. The benchmark spans 25 programs across Unix utilities, bioinformatics, interpreters, cryptography, compression, and more.

​We’ll discuss what MirrorCode reveals about current AI coding capabilities, how to evaluate weeks-long autonomous work, and the methodological challenges of measuring progress on real software projects.

​Learn more: https://epoch.ai/MirrorCode

​Schedule
7:00pm - Doors open, light refreshments
7:30pm - talk begins

​Check out our other Summer Season events here → moxsf.com/summerseason