2 months ago
Microsoft's Copilot is getting moreover amended astatine investigation acknowledgment to a caller characteristic that combines nan powerfulness of some OpenAI's ChatGPT and Anthropic's Claude. In a blog station announcing Copilot Cowork's availability, Microsoft debuted nan Critique characteristic that will beryllium utilized successful Microsoft 356 Copilot's Researcher tool. Unlike nan modular Copilot, Researcher is designed to tackle much analyzable tasks pinch aggregate steps.
Now, Researcher is getting moreover amended astatine that pinch nan Critique characteristic that uses GPT responses, which are past refined by Claude. In a blog post, Microsoft said that, "this architecture creates a powerful feedback loop that delivers higher-quality results crossed actual accuracy, analytical breadth, and presentation," adding that Researcher's process is akin to what you spot successful "academic and master investigation settings." Microsoft claims nan upgrade scores somewhat higher (compared to nan astir caller Perplexity Deep Research models) connected nan Deep Research Accuracy, Completeness, and Objectivity benchmark. On its own, Anthropic has a Research feature that tin usage aggregate Claude agents to supply a broad consequence to much analyzable requests.
If you for illustration doing investigation pinch a small much autonomy, Microsoft besides added nan Model Council characteristic that's disposable arsenic an replacement action for Researcher. With Model Council, you'll get side-by-side responses from some Anthropic and OpenAI, pinch a study that shows wherever nan models work together and disagree. Both features are presently disposable successful Microsoft 365 Copilot's Frontier program, which acts arsenic a early entree abstraction for nan company's AI innovations.
