THREE THINGS TO TAKE WITH YOU
- Match each example with its actual evidence type.
- Notice the decisions and expertise supplied by people.
- Adapt a process at a scale you can try.
A collection of possibilities
The supplied casebook gathers human–AI experiences across education, accessibility, art, research, and community science. This edition selects examples that readers can explore through project documentation and pairs them with ordinary, achievable activities.
The selection is curated. It is not a statistical picture of all AI use, and the different forms of evidence are not interchangeable. A completed artwork shows that a particular work was made. A project description explains a design. A research paper describes a study with its own methods and setting.
Learning through dialogue
The casebook presents Harvard's CS50 Duck as an example of a course team designing AI support around hints and explanation. The primary Teaching CS50 with AI paper documents the team's educational work and deployment. In this edition, it is labeled a teaching project rather than using participation counts as proof of learning improvement.
Our practical adaptation is to ask for one guiding question at a time while explaining a concept in your own words. Save the explanation before and after the exchange. The difference is a useful artifact to reflect on.
Access as an ongoing conversation
Be My AI appears in the casebook as an example of image description with follow-up dialogue. The user's question helps determine which visual detail is relevant. The interesting pattern is not just “describe an image,” but “let the person ask the next question about what matters to them.”
A workshop adaptation is to offer a written description of shared visual material and invite participants to request the details that would help them participate. That is a facilitation practice, not a claim that this website supplies an image-description service.
Creative instruments and human choices
The casebook's artistic examples include Holly Herndon's Holly+ and Sougwen Chung's Drawing Operations. They make different aspects of collaboration visible: a voice model offered as an artistic instrument, and a physical drawing practice developed with robotic systems.
For a small exercise, name the kind of contribution you want from a tool. Ask for a variation in rhythm, arrangement, or language. Make the selection yourself, and record the choices that give the work its identity. A useful artist statement describes the intention and process rather than only listing software.
Observation, computation, and shared inquiry
The casebook includes iNaturalist's combination of field observation, computer-vision suggestions, and community identification. It also discusses Foldit's use of players' spatial reasoning with computational modeling. These are different technical systems; Foldit is described here as human–computation collaboration rather than relabeled as a modern generative-AI chatbot.
The shared invitation is to contribute an observation or a piece of reasoning to a larger inquiry. Begin by describing what you actually saw or tried, then consult the relevant community or source. Keep the contribution and its context together.
Design and scientific models
To connect the casebook with the other supplied reports, the public collection also includes JPL's generative-design lander concept and AlphaFold 3's biomolecular structure-prediction research. These are labeled according to what they demonstrate: a design concept and a research model.
A scientific model is a way to investigate a question. A design concept is a way to explore possible structures. The practical lesson for a small creative project is to make alternatives inspectable, choose a useful comparison, and document how people interpreted the result.
Read a case, make a turn
Every case page separates the goal, human contribution, computational contribution, documented output, and a proposed activity. The source link lets you continue to the original account.
Choose one case because of the process it illustrates, not because it is the largest or most impressive. A guiding question, a better description, a new arrangement, or a carefully documented observation may be exactly the next turn your project needs.
Sources & further reading
Source basis for this edited guide · Positive Human-AI Casebook Research.md
The supplied report is the basis for the synthesis. Practice examples and interfaces are editorial adaptations. Selected primary-source checks are labeled below; other links are further reading, not a claim of independent verification.
- Teaching CS50 with AI — Harvard ↗Primary paper checked
- Introducing Be My AI — Be My Eyes ↗Primary project account checked
- Research — Sougwen Chung ↗Primary artist documentation checked