Reddit answers Launches Chatbot in France, Immediately Faces Controversy Over Harmful Advice
Paris, France – Reddit Answers, the discussion forum’s new chatbot powered by Google’s Gemini model, has launched in France amid growing concerns after reportedly suggesting heroin and kratom as potential remedies for pain. The incident highlights the risks of relying on AI trained on user-generated content, even as Reddit profits from data licensing agreements with major tech companies like Google and OpenAI.
The chatbot, accessible at the bottom of the screen, is intended to provide users with answers based on the vast trove of discussions within the Reddit platform. However, a user recently discovered the AI recommended illicit drugs when seeking details about pain relief. While Reddit has since adjusted the chatbot’s programming to avoid answering questions about heroin, the initial lapse raises questions about the safety measures in place for AI tools learning from possibly unreliable sources.
Reddit’s platform is a valuable resource for artificial intelligence developers, with a reported $60 million annual agreement with Google to utilize its data for model training, and a similar arrangement believed to be in place with OpenAI. The incident underscores the inherent challenges of training AI on unfiltered forum discussions,where misinformation and harmful advice can proliferate.
The controversy arrives as Reddit seeks to expand its AI-driven features, offering a new way for users to navigate the platform’s extensive content. The company has not yet commented on the specific incident, but the situation is highly likely to fuel debate about the responsible growth and deployment of AI chatbots, particularly those drawing information from public forums.