In a newly published article, Microsoft introduces a new feature called Critique, integrated into Microsoft 365 Copilot's Researcher tool. This tool is designed to handle complex, multi-step tasks, instead of just simple answers.
With the new feature, the initial answer from OpenAI's system will be checked and revised by Anthropic's system. Microsoft said this approach creates a continuous feedback loop, helping to improve content quality in terms of accuracy, level of analysis and presentation. This process is similar to research in academia and professional environments.
According to Microsoft, the upgraded version gives better results in accuracy, completeness and objectivity compared to some new AI research systems of Perplexity AI.
In addition, Microsoft also added the Model Council feature. This feature displays parallel answers from two different AI systems, while pointing out similarities and differences for users to self-assess.
Microsoft Copilot is an artificial intelligence assistant developed by Microsoft, integrated into many applications such as Word, Excel, Outlook or Windows, to support users in writing content, analyzing data and searching for information.
Unlike a single model, Copilot operates as a central platform, which can connect and coordinate many different AI systems.
