An Investigation of Human Annotators' AI Teammate Selection and Compliance Behaviours
Bibliographic record
Abstract
Human-artificial intelligence (AI) collaborative annotation has gained increasing prominence as a result of its enormous potential to complement human and AI strengths as well as AI's recent development. However, it is not straightforward to form suitable human-AI teams and design human-AI interaction mechanisms for effective collaborative annotation. Through an exploratory study, this thesis investigated a diverse set of factors that may influence humans' AI teammate selection and compliance behaviours in a collaborative annotation context wherein AI agents serve as suggesters to humans. The study results indicate that multiple factors influenced which AI agents the participants chose to receive suggestions from, such as the AI agents' recent and overall accuracies as well as the participants' suggestion compliance records. We also discovered that the participants' AI compliance decisions were swayed by factors including whether the AI agents' suggestions aligned with the participants' top choices and whether such suggestions provided novel perspectives to the participants. Moreover, it was found that most of the participants constructed narratives to interpret the differences in various AI teammates' behaviours based on limited evidence. This thesis also contributes by presenting MIA, a versatile web platform for mixed-initiative annotation. Based on the weaknesses of MIA's current designs, as informed by empirical results of the aforementioned exploratory study and another human-AI collaborative annotation study, as well as the goals to improve MIA's scalability and adaptability, this thesis proposes design changes to MIA; these design changes also apply to other mixed-initiative annotation platforms.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.040 | 0.137 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.004 | 0.002 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.001 | 0.004 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".