Microsoft Debuts Critique Multi-Model System and Council Tool to Boost Copilot Accuracy
Microsoft introduced Critique, a multi-model AI research system for Copilot that generates, reviews and refines responses through cross-checking several models. It also launched Council, a side-by-side comparison tool using a judge model to highlight differences and surface top insights, enhancing verification to address AI inaccuracy concerns.
1. Launch of Critique for Copilot
Microsoft rolled out Critique, a deep research system that orchestrates several AI models to generate, review and refine responses for Copilot. This multi-model approach cross-checks outputs to enhance accuracy and reduce the likelihood of confident but incorrect answers.
2. Introduction of Council comparison tool
The new Council feature produces side-by-side reports of different model responses and employs a separate judge model to highlight distinctions, surface unique insights and summarize the strongest points. This tool aims to give users a clearer view of model trade-offs and decision logic.
3. Strategic impact on Microsoft’s AI offering
Building review and verification into Copilot underscores Microsoft’s push to establish AI as a serious productivity platform for research, writing and decision-making. Investors will be watching if these enhancements drive enterprise adoption and differentiate Microsoft in the intensifying AI competition.