OpenAI has reportedly hired hundreds of contract workers to review user conversations with ChatGPT as part of an internal initiative codenamed "Project Lily." These workers evaluate conversations in three stages: reading the prompt, summarizing the user's intent, and then rating the model's response. While efforts are made to anonymize data, some conversations contained sensitive personal information, and OpenAI acknowledged that complete privacy protection is challenging. Anthropic has also admitted to using similar human review processes for its models. AI
IMPACT Highlights the ongoing challenges and methods used in AI model training, particularly concerning data privacy and human oversight.
RANK_REASON News about a company's internal process for improving a product, not a new product release or core research.
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
