Google Deploys New AI Model to Improve Accuracy of Featured Snippets
Google is rolling out an AI update that cross-references featured snippet answers with other top search results to ensure accuracy. The new system also reduces snippets for nonsensical questions by 40 percent.
Google is updating its featured snippets with a new artificial intelligence system called the Multitask Unified Model. This AI model improves the quality of the top info boxes by cross-referencing the bolded snippet text against other high-quality search results to verify that they convey the same information, even if the wording differs.
The new system also tackles the issue of flawed search queries by recognizing when a question lacks a logical premise. As a result, Google now displays featured snippets 40 percent less often for nonsensical searches, preventing misleading answers for questions like asking when a cartoon dog assassinated a historical figure.
In addition to these snippet improvements, Google is expanding its use of warning messages for searches where the engine lacks high confidence in the available results. While these content advisories previously appeared mainly for emerging topics, the search giant now deploys them in additional situations to better protect users from unreliable information.