Home / Technology / Microsoft Copilot's New 'Critique' Feature Boosts AI Research
Microsoft Copilot's New 'Critique' Feature Boosts AI Research
30 Mar
Summary
- New feature combines ChatGPT and Claude for advanced research tasks.
- Critique feature uses AI feedback loop for superior accuracy and breadth.
- Model Council offers side-by-side AI responses for comparison.
Microsoft's Copilot is significantly advancing its research capabilities with a new feature called Critique. This innovation integrates the strengths of both OpenAI's ChatGPT and Anthropic's Claude within the Researcher tool, designed for complex, multi-step tasks.
The Critique feature establishes a powerful feedback loop, where GPT responses are refined by Claude. Microsoft states this architecture enhances factual accuracy, analytical breadth, and presentation quality, mirroring professional research methodologies. It reportedly outperforms recent Perplexity Deep Research models on key benchmarks.
Alternatively, Microsoft offers the Model Council feature for users who prefer more autonomy. This option presents side-by-side responses from both Anthropic and OpenAI, detailing model consensus and divergence. These advanced research tools are currently available in the Microsoft 365 Copilot Frontier program.
