Personalized Auto-Research: Towards a True AI Co-Scientist
The paper argues that AI research agents need to model the individual scientist, not just the task.
The authors call this “personalized auto-research”: carrying a researcher’s prior work, methods, collaborators, and communities through hypothesis generation, retrieval, experiments, writing, and review. They frame generic systems as a failure mode, because two researchers asking the same question may get essentially the same output. Their proposed framework centers on graph-grounded researcher representations and evaluation tied to the individual user. HF Daily Papers' note
The authors call this “personalized auto-research”: carrying a researcher’s prior work, methods, collaborators, and communities through hypothesis generation, retrieval, experiments, writing, and review. They frame generic systems as a failure mode, because two researchers asking the same question may get essentially the same output. Their proposed framework centers on graph-grounded researcher representations and evaluation tied to the individual user. HF Daily Papers' note
score 4