Google Faces Criticism Over New Guidelines for Gemini AI Fact-Checking

Contractors evaluating Gemini's responses are now required to assess prompts outside their expertise, raising concerns about AI accuracy.

Google has reportedly updated its guidelines for contractors evaluating Gemini AI, removing the option to skip prompts outside their expertise.
Contractors working with GlobalLogic, an outsourcing firm, are now instructed to rate prompts they partially understand and note their lack of domain knowledge.
Previously, contractors could skip prompts requiring specialized knowledge, such as technical or medical queries, to ensure more accurate evaluations.
Critics argue the new policy could compromise Gemini's accuracy, particularly on complex or sensitive topics like healthcare and coding.
Google has not responded to media inquiries regarding the updated guidelines or the concerns raised by contractors and analysts.