Google is expanding its efforts to improve the evidence base of AI chatbots for businesses by integrating real financial data from business and financial services.
The plans aim to reduce inaccuracies in AI-generated information and improve overall reliability, known as “hallucinations,” by basing responses on verified data.
The news is part of Google’s plans to add external data sources to Web search engines and internal company data. The first partners are Moody’s, Thomson Reuters and ZoomInfo.
Google wants enterprise AI to be more factually accurate
Late to the party, companies had initially been waiting for more secure and private systems to protect sensitive corporate data, but as more companies deploy AI solutions, the focus has shifted to accuracy.
Speaking about the changes, Google Cloud CEO Thomas Kurian said Axios: You can actually trust the model to perform a task on your behalf because you have a basis for trusting it.”
To further improve reliability, Google is also introducing a trust score, which provides a numerical indicator of how confident the AI model is in its response. Enterprise users can also instruct the AI chatbot to prioritize information from specific documents or data in a prompt, rather than the broader training data.
Kurian added: “We taught the model how to ensure that when responding, it treats the information in the input prompt as the primary information it should pay attention to.”
Additionally, Google is expanding Vector Search to support hybrid searches, which combine vector-based searches with text-based keyword searches for greater accuracy. The upgrade is currently in public preview.
Google Clouds announcementwritten by Burak Gokturk, VP & GM for Cloud AI & Industry Solutions, concludes: “As these technologies become even more capable, we want to help enterprises realize the full potential of grounded generative AI in the real world.”