AI for One Welfare: the role of animal welfare scientists in developing valid and ethical AI-based welfare assessment tools
Bibliographic record
Abstract
The increasing use of artificial intelligence (AI) in livestock farming is accelerating the development of automated welfare assessment tools, particularly with advancement in generative AI such as large multimodal models (LMMs). Yet, animal welfare scientists have rarely been involved in the development process of these tools or their subsequent adaptation within the field. Here, we discuss possible roles for animal welfare scientists in the development and validation of AI-based welfare assessment tools. We first examine key uncertainties that emerge during development, including the selection of relevant, valid and reliable welfare indicators and gold standards, hardware and software solutions for data collection, methods for integrating multiple welfare indicators, and the real-world impact of automated welfare assessment tools. Second, we demonstrate the use of LMMs to assess welfare based on a case study using dairy cow cleanliness. Finally, we consider the practical implementation of AI-based welfare assessment and discuss potential tensions around (1) embedded values in LMMs, (2) AI's influence on decision-making on farms, (3) the integration of AI in current knowledge systems by human-AI collaboration, and (4) the economics of AI-based welfare assessment and improvement. We conclude that LMMs could help automate welfare assessment and communicate results to humans in accessible formats, but outcomes depend on which stakeholders are involved in the development process. We advocate for developing AI-based welfare assessment tools through the One Welfare framework, recognizing that AI deployment affects humans, animals, and the environment simultaneously, and suggest potential pathways for animal welfare scientists to engage in the process.
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 | 0.175 | 0.162 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.006 | 0.003 |
| Science and technology studies | 0.006 | 0.040 |
| Scholarly communication | 0.021 | 0.019 |
| Open science | 0.006 | 0.016 |
| Research integrity | 0.007 | 0.013 |
| 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".