ChatGPT Adds Reading Feature to Enhance User Interaction Experience


Recent exposure reveals OpenAI's internal "Project Lily" uses human prompt reviewers to manually review and score real ChatGPT user chats. This challenges the belief that large model evolution relies only on algorithm upgrades and automated data, showing human review remains key in training.....
OpenAI has been revealed to have hired hundreds of contract workers to review real ChatGPT user conversations, rating responses on a scale from 1 to 7 to improve the model, with a focus on reducing excessive flattery and anthropomorphism. Many users are unaware that their chats may be read by real people; anonymization can still contain sensitive personal information, and OpenAI's privacy filter has also been acknowledged to potentially have vulnerabilities.
At the 2026 Bund Conference, Ant Group CEO Han Xinyi said the 'ChatGPT moment' for agent commerce has arrived. Agents are moving from capability demos to real transactions, becoming the core entry linking user demand and supply. After ChatGPT's natural language leap, agents now deepen the business loop.....
At the 9th September Bund Conference in Shanghai, Han Xinyi, CEO of Ant Group, stated that the 'ChatGPT moment' for intelligent agent business has arrived. The core shift is from 'capability demonstration' to 'real transactions'. Intelligent agents are no longer just Q&A generation tools, but new commercial entry points that understand user needs, connect supply, and drive transactions. User demand is shifting from consultation to transactions.
OpenAI Korea's GM says collaboration with Samsung Electronics on next-gen chip co-production and R&D has achieved significant results. Samsung is among the largest global ChatGPT users, widely applying it in R&D, marketing, and sales. Both are advancing faster chip technology development.....