All Image Captions Produced are Violent
April 2, 2018
The model was trained on a dataset that includes violent and disturbing content, which has been linked to an increase in aggression and hate speech online. As a result, the generated captions may perpetuate these harmful ideologies.
The release of Norman highlights the disturbing implications of training AI models on violent and hateful content, which can contribute to the spread of toxic ideologies online. As we move forward in developing AI technology, it is crucial that we take a closer look at the data we provide to these models, ensuring that they do not perpetuate harm. By joining together to demand greater accountability from tech companies, we can work towards creating a safer digital landscape and preventing the spread of violent image captions. And to stay informed about AI incidents, visit JOIN US.
Matched TAIM controls
Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls
- MAP 4.2 — similarity 0.597, rank 1. TAIM detail and related incidents →
- MAP 2.1 — similarity 0.585, rank 2. TAIM detail and related incidents →
- GOVERN 4.2 — similarity 0.584, rank 3. TAIM detail and related incidents →
- Alleged deployer
- MIT Media Lab
- Alleged developer
- MIT Media Lab
- Alleged harmed parties
- Unknown
AI governance case studies
For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.
Source
Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/41
Data source
Incident data is from the AI Incident Database (AIID).
When citing the database as a whole, please use:
McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.
Pre-print on arXiv · Database snapshots & citation guide
We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.