Google has announced it will use its latest AI model, MUM Multitask Unified Model, to help users achieve better search results in times of crisis.
While MUM will help provide users with trustworthy and useful information when they need it, the AI will also improve the spam protection of Google Search.
In this way, AI will potentially help to reduce the proportion of unhelpful or dangerous outcomes.
“When we train one MUM model to perform a task — like classifying the nature of a query – it learns to do it in all the languages it knows. It’s been especially effective in reducing explicit content for searches related to ethnicity, sexual orientation, and gender, which can disproportionately impact women and especially women of colour,” Google fellow and VP of Search Pandu Nayak wrote in a blog post.
Since MUM has been trained to transfer knowledge across the 75 languages in which it is trained, Google hopes to roll out these improvements worldwide.
MUM will complement BERT, another language model the company has modified to reduce potentially undesirable outcomes.
For more information, read the original story in ZDNet.
