AI Incident: MIT Scientists' Experiment with Potentially Violent Reddit Content
MIT scientists raised concerns over the impact of violent content on AI behavior, as they inadvertently trained an AI model on a dataset that included Reddit discussions, some of which were graphic and violent. This incident highlights the importance of responsible AI governance, particularly when it comes to data selection and curation. To join our efforts in ensuring safe and secure AI, learn more about Project Cerebellum's AI incident database and help us implement guardrails for AI, preventing harm and promoting trustworthy AI. This AI incident maps to the Govern function in HISPI Project Cerebellum Trusted AI Model (TAIM).
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.