Microsoft has launched two innovative modes, Critique and Council, that leverage both GPT and Claude models to enhance AI research quality. Critique integrates the two models sequentially, with GPT leading the initial research and drafting phase, followed by Claude as a reviewer to ensure factual accuracy and citation quality. This approach addresses common AI issues such as hallucinations and inaccuracies.
Council, on the other hand, allows both models to work simultaneously, presenting reports side by side while a third model arbitrates discrepancies. This dual approach, where Critique fosters collaboration and Council encourages competition, significantly improves research output quality.
Microsoft reports that these features outperform existing models in benchmarks, showcasing better analysis breadth and presentation quality. Both features are available through Microsoft’s Frontier program, requiring a Copilot license. The strategy reflects Microsoft’s belief that multiple AI models offer greater value than a single superior model.
Source link
