AI agents reimplement HCI research code from papers
Researchers use AI agents to 'revibe' HCI systems directly from papers
Software artifacts for most technical HCI research projects are unavailable, limiting academic knowledge production. Researchers demonstrate that agentic AI can "revibe" interactive software by reimplementing systems directly from research papers. They describe a revibeability metric and test the approach by revibing recent UIST papers and interviewing the original authors. The results are encouraging—in many cases, the reimplemented code is suitable for strong baseline use. The authors argue this could fundamentally shift how the technical HCI community produces, uses, and evaluates research artifacts.
- AI agents can reimplement interactive HCI systems directly from research papers using a new 'revibability metric'
- Researchers recreated systems from recent UIST papers, achieving baselines suitable for follow-up research
- This approach could address the reproducibility crisis in HCI by automating the creation of research artifacts
Why It Matters
AI-driven code revival could eliminate a major bottleneck in HCI research and accelerate innovation.