Google Unveils MUM AI to Tackle Complex Multi-Step Searches
Google introduces MUM, a powerful new AI model that is 1,000 times stronger than BERT and designed to understand complex search tasks requiring multiple queries. The multimodal technology aims to provide expert-level answers across 75 languages to save users time.
Google introduces a new artificial intelligence technology called Multitask Unified Model, or MUM, to solve complex search tasks that currently require multiple queries. People typically issue eight separate searches for complicated questions, such as comparing the preparation needed to hike different mountains, but MUM allows users to ask a single comprehensive question instead.
This new AI model is 1,000 times more powerful than the previous BERT framework and demonstrates advanced capabilities by both understanding and generating language. Google trains MUM across 75 different languages and various tasks simultaneously, which gives the system a much deeper understanding of information and world knowledge than earlier models.
Unlike previous technologies, MUM is multimodal and processes information across text and images, with future plans to include video and audio. By understanding the nuanced context of complex tasks just like a human expert would, MUM ultimately reduces the number of searches people need to accomplish their goals.