Google is launching a new initiative aimed at enhancing the reliability of generative AI by leveraging its search engine to provide grounding for AI-generated content. Part of a broader set of updates at Cloud Next, the update is designed to mitigate the issue of “hallucinations” or inaccuracies in AI outputs by offering users more current information and verified sources.

Google also announced that it’s integrating its search capabilities into Vertex (its AI solution for businesses), allowing for the use of both public internet data and proprietary company data to improve AI query results. This move is in line with similar efforts by Microsoft with its Bing search engine and Copilot chatbot, underscoring a growing industry trend towards enhancing AI reliability through grounding.

Limiting hallucinations for LLMs is a good goal, unless you need the probabilistic output to enhance the outcome you are seeking. Said differently, sometimes a little nonsense is a feature, not a bug. Generally speaking, however, the more accurate and deterministic the output, the better for most business use cases.

More By This Author: