# Family name: what the field calls a person, a model and a machine on one project

A first pass of 27 sources. It finds that the field's names are intelligence augmentation and hybrid intelligence, and that this family fits augmentation better than teaming.

## Files

| File | Role |
|---|---|
| [frame](frame.md) | step 1: the problem and the field's names for it |
| [protocol](protocol.md) | how the pool was built, and what was not searched |
| [sources](sources.md) | the flow, then each work with its grade or score |
| [review](review.md) | step 2: surveys, seminal works, practice |
| [state of the art](state-of-the-art.md) | step 3: best evidence beside the practice most adopted |
| [position](position.md) | step 4: which field, and what the family adds |
| [against](against.md) | where the combination fails |
| [baseline](baseline.md) | step 5: the named baseline, H1 |
| [bibliography](../bibliography.md) | every work cited, once |

## Claims

| Claim | Grade | Where |
|---|---|---|
| The field's names are intelligence augmentation, hybrid intelligence, human-AI teaming and human-in-the-loop | convention, from surveys and definitions | frame.md |
| Human-AI combinations did worse than the best of either alone on average, g of -0.23 (secondary) | measured | against.md |
| A model raised output on writing and coding tasks inside its reach (secondary) | measured | state-of-the-art.md |
| The checking machine is not a third role in the literature read | supported with caveat: one reviewer, a small pool | position.md |
| Agent files and agent patterns are adopted without a controlled study | convention, untested | review.md |