A human-centered index of human-AI coevolution.

The Human–AI Coevolution is a curated, openly maintained list of research on how humans must evolve to use AI well as AI advances. As humans delegate more to AI, the capabilities humans need on their own side change — from critical thinking, to evaluative expertise, to metacognitive monitoring, to systems thinking. This index curates the research that documents that human evolution. The site is a presentation layer: every paper, link, author, and tag lives in papers.yaml.

§01 Scope

The canonical index includes papers that bear on how humans develop and sustain the capabilities needed to use AI well at each phase. Many of these papers also document how humans' choices shape future AI — for instance, how feedback signals during use become training data — but the framing is human-first: human capability is the dependent variable; AI is the environment that's changing around it. One-way capability papers and one-shot UX studies belong elsewhere.

Papers that materially inform the area but don't centre the human side — foundational RLHF method papers without a coevolutionary lens, single-direction behavioural studies later cited by this work, related theoretical frameworks — go into adjacent papers. Adjacent entries are excluded from the main browse, stats, and per-category groupings so the canonical signal stays scoped.

§02 Categories

Each paper carries one or more category tags. The 2-letter codes (CC, MA, HF, LH, PS) are used as compact labels throughout the site.

CC

Collaboration & Co-Creation Humans and AI working jointly on a task — how the human's strategy and the AI's outputs reshape each other within and across sessions.

MA

Mutual Adaptation Both human and AI behaviour measurably change in response to each other over time.

HF

Human Feedback Loops RLHF, preference learning, training-from-use — with explicit treatment of how the trained model reshapes future human inputs.

LH

Longitudinal HCI Studies Real-world deployments over weeks or months measuring shifts in human behaviour, skill, and norms around AI.

PS

Position & Survey Framing pieces, surveys, and theoretical frameworks that argue for or characterize the coevolutionary lens.

§03 Adding a paper

Each entry is a YAML object with a fixed structure: title · link · authors · institutions · date · publisher · category · keywords · TLDR. The repository's contribution guide documents the canonical formatting rules; a local pipeline regenerates the README, per-category and per-keyword groupings, and the statistics figures whenever the source changes.

For dates, we always use the earliest known public release (arXiv v1 over the camera-ready), so the date reflects the work's true age.

§04 Contribute

• 01 Suggest a paper — open an issue with the title, link, and any relevant details.
• 02 Fix metadata — every paper entry has an "Edit on GitHub" link that goes to the entry's exact line in the YAML source.
• 03 Submit directly — edit papers.yaml, run bash scripts/update_repo.sh, and open a PR with the regenerated diff.

Some of the design and scaffolding here is adapted from OSU-NLP-Group/GUI-Agents-Paper-List. Thanks for their awesome work!

If this list helps your work, please consider starring the repo — it's the simplest signal that helps others find it.