Microsoft asks Claude to keep ChatGPT honest in new Copilot feature
Similar to "academic and professional research" processes, says Microsoft of defaulting to two-engine research mode.
Microsoft pitched a new form of Copilot use, promoted by CEO Satya Nadella on Monday, that pits Anthropic and OpenAI's models against one another.
Users in the frontier program for Microsoft 365 Copilot can now use "critique" and "council" modes in the research agent somewhat predictably named Researcher, Microsoft said.
Critique is the default mode, and responses will be generated by ChatGPT and then refined by Claude. (The reverse, with ChatGPT moderating Claude, is not yet available.)