Microsoft has unveiled advanced AI features for its Copilot assistant, enhancing its enterprise AI tools. The upgrade includes a new “Critique” feature in the Researcher tool, integrating OpenAI’s GPT with Anthropic’s Claude for improved reliability. GPT generates content while Claude verifies its accuracy, significantly reducing AI hallucinations. Nicole Herskowitz highlighted the benefit of utilizing multiple models, which elevates productivity and quality. Additionally, the “Council” feature presents outputs from various AI models side by side, summarizing areas of consensus and disagreement, reflecting an academic-like evaluation process. Internal tests revealed that this multi-model system significantly surpassed single-model approaches, with a 13.88% improvement in aggregate scores. These innovations are part of Microsoft’s Frontier program, responding to the rising demand for autonomous AI agents. In parallel, Microsoft plans to invest $1 billion in Thailand’s AI infrastructure and digital skills development, addressing increasing competition from key players like Google and Anthropic.
Source link
