Will using artificial intelligence to review camera trap images reduce human connection to wildlife research?
Bibliographic record
Abstract
Abstract Camera traps have been widely adopted in wildlife research and management; however, the manual review of the large volumes of images they produce is time‐consuming and prone to errors. Artificial Intelligence (AI) platforms are becoming increasingly popular for automated image processing, as these tools can significantly reduce the time needed for reviewing images. Given the need for high‐quality, rapidly accessible data for implementing conservation actions in a changing world, the use of AI holds great promise for conservation. Despite the potential of AI, we raise concerns regarding the loss of the human element in wildlife data review. We argue AI may miss unexpected discoveries in images and diminish the personal connection to wildlife and conservation landscapes which is fostered through manual image review. As human values are pivotal in soliciting investment in conservation, AI may pose a risk through the loss of human connection to ecological systems. Further, outsourcing image review to AI represents a loss of training opportunities for the next generation of scientists. Manual review of images also engages citizen scientists in scientific discoveries, fostering enthusiasm for conservation careers and community support for conservation actions. While acknowledging the benefits of AI in processing wildlife camera trap images, we call for meaningful conversation on how AI should be used in the advancement of wildlife research. Given the recent challenges our field has faced with the advent of large language models (e.g., ChatGPT) in scientific training and research production, we should proactively begin conversations. We should prepare for discussion on alternative means of maintaining human connection to wildlife research, and alternative training opportunities for students and citizen scientists.
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 | 0.102 | 0.483 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.004 | 0.006 |
| Scholarly communication | 0.011 | 0.017 |
| Open science | 0.004 | 0.005 |
| Research integrity | 0.004 | 0.004 |
| Insufficient payload (model declined to judge) | 0.017 | 0.005 |
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".