Google Joins Forces with Thomson Reuters, Moody’s, and Others to Equip AI with Real-World Data

As Google aims to attract enterprise customers, it is working to reduce hallucinations in its AI platform, a critical concern for organizations that are already cautious about adopting this technology. To bolster confidence, Google is enhancing model grounding by partnering with reputable third-party services: Moody’s, MSCI, Thomson Reuters, and Zoominfo. These partnerships will be integrated into Vertex AI starting next quarter, providing developers with reliable data that supports accurate model outputs.

For enterprise developers, this means they can utilize high-quality, vetted data and the expertise of Subject Matter Experts to meet their standards. As Nick Reed, Moody’s Chief Product Officer, stated in a recent press briefing, "Our data will be available to others so they can ground their responses for internal or external interactions. Google's service allows clients to leverage both our data and their models, integrating them into applications that enhance user experience."

Google also incorporates its Search capabilities into this grounding service, enabling businesses to supplement Gemini outputs with up-to-date, high-quality web information. This offers companies a more reliable source of truth, avoiding the pitfalls of potentially misleading blog content that may manipulate SEO for authority.

Moreover, Google is introducing high-fidelity grounding through an experimental preview. This feature is designed to enhance AI systems’ performance with specific information sets, particularly useful for tasks like summarizing multiple documents or extracting key details from financial reports. Powered by Google’s Gemini 1.5 Flash, high-fidelity grounding ensures that AI focuses on relevant prompts instead of being influenced by unrelated training data.

When asked if these new partnerships and features establish Google's generative AI as the most reliable model compared to competitors, Google Cloud Chief Executive Thomas Kurian highlighted three differentiators: First, Google leverages its reputation as a trusted source for web data and real-world insights through Google Search. Second, it provides customization options for data grounding in specific sub-verticals—giving users control over their grounding criteria. Finally, high-fidelity grounding focuses AI responses on the most relevant information, thus enhancing overall quality.

"These three elements together offer the highest level of control over the quality of responses generated by the model," Kurian explained. "This is part of our commitment to reducing hallucinations and increasing trust in AI models."

This announcement coincides with other significant updates from Google Cloud, including the public release of Gemini 1.5 Flash and 1.5 Pro, Gemma 2, and additional Vertex AI enhancements.