Navigating accountability: the role of paradata in AI documentation and governance
Bibliographic record
Abstract
Purpose The increased use of Artificial Intelligence (AI) has prompted governments internationally to provide guidance and legislation to maximize the benefits of AI while minimizing the risks to humans and organizations. This paper explores how published requirements for documentation in a sampling of authoritative texts address the challenges of creating, capturing and preserving records of the design, implementation and use of AI tools for accountability and transparency, and how the analytical concept of paradata can help to meet the recordkeeping challenges presented by the design, development and implementation of AI systems. Design/methodology/approach Inductive reading and conceptual analysis of a set of AI laws, regulations and frameworks published by the EU, UK, USA, Canada and Singapore. Findings The authoritative texts reviewed clearly describe activities which imply the necessity of records creation and preservation. Identifying specific documents necessary to comprise a sufficient body of records to provide evidence of accountable AI implementation and operation can be difficult. Literature on paradata in archival applications of AI may prove productive in identifying relevant information artifacts for preservation in the AI process. Paradata is produced by those designing and implementing AI systems and by AI systems themselves. Practical implications Identifying relevant paradata produced by AI systems requires archivists to develop both the capacity to analyze and the vocabulary to discuss these systems in order to preserve evidence of their operation in compliance with legislation and international standards. Originality/value No comparable comparative analyses have been published in the archives and information field.
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 | 0.136 | 0.246 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.012 | 0.013 |
| Science and technology studies | 0.018 | 0.060 |
| Scholarly communication | 0.031 | 0.035 |
| Open science | 0.003 | 0.016 |
| Research integrity | 0.003 | 0.005 |
| Insufficient payload (model declined to judge) | 0.005 | 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".