Microsoft combines claude and chatgpt for enhanced ai research

Microsoft has announced the launch of Copilot Cowork, an AI tool that leverages the capabilities of both Claude from Anthropic and ChatGPT. This strategic move aims to improve the accuracy and completeness of research outputs.

Enhanced researcher functionality

The Researcher tool within Microsoft 365 has received a significant upgrade with the introduction of a new Critique function. This feature utilizes responses from both ChatGPT and Claude, allowing for a deeper analysis and evaluation of the information provided.

According to Microsoft, this combination of models from leading labs like Anthropic and OpenAI results in a 'dual verification' process, reducing potential errors. The workflow now involves a model planning the task, generating an initial draft, and a second model refining the response, acting as an expert reviewer before providing the final output.

This approach yields more comprehensive and accurate results, with Researcher now boasting a 13.8% superior performance in the DRACO (Deep Research Accuracy, Completeness, and Objectivity) benchmark, a standard in the industry.

Council for comparative analysis

Microsoft has also unveiled Council, a tool allowing users to compare the responses of ChatGPT and Claude side-by-side. This feature provides a broader perspective, enabling users to assess the similarities and differences between the two models' outputs and the value each brings to the table.

Automated comparisons are made possible, offering an even more streamlined experience. With Copilot Cowork now available on Microsoft's Frontier program, users can harness the power of this combined AI offering for enhanced research capabilities.