Entities
Allen Institute for AI
Incidents involved as both Developer and Deployer
Incident 1463 Reports
Research Prototype AI, Delphi, Reportedly Gave Racially Biased Answers on Ethics
2021-10-22
A publicly accessible research model that was trained via Reddit threads showed racially biased advice on moral dilemmas, allegedly demonstrating limitations of language-based models trained on moral judgments.
MoreRelated Entities
Other entities that are related to the same incident. For example, if the developer of an incident is this entity but the deployer is another entity, they are marked as related entities.
Related Entities
Other entities that are related to the same incident. For example, if the developer of an incident is this entity but the deployer is another entity, they are marked as related entities.