Microsoft Unveils Critique: AI Duo from OpenAI and Anthropic Boosts Research Precision in Microsoft 365 Copilot

Microsoft has introduced an update to its Researcher tool within Microsoft 365 Copilot, which allows simultaneous use of multiple artificial intelligence models. It refers to the combination of ChatGPT from OpenAI and Claude from Anthropic.

What is Known

The new Critique feature operates on sequential processing principles. Initially, a response is generated by the GPT model, after which Claude analyzes and refines it. Microsoft notes that such an architecture creates a "powerful feedback loop," enhancing factual accuracy, breadth of analysis, and presentation quality. According to the company, this approach is akin to academic and professional research processes.

The Researcher tool is geared toward complex multi-step tasks. Unlike the standard Copilot, it is better suited for deep analysis. Thanks to the new approach, the system shows higher results in terms of accuracy, completeness, and objectivity compared to deep search models from Perplexity.

In addition, Microsoft launched an alternative feature, Model Council. It allows for receiving answers simultaneously from two models, OpenAI and Anthropic, with concurrent display of their conclusions and discrepancies as a separate report.

Both functions are currently available within the Frontier program, which serves as an early access environment for new artificial intelligence developments.

Source: Microsoft