Google Assistant is reportedly making a strategic shift towards generative AI, as mentioned in an internal email reported by Axios. The Assistant team sees immense potential in exploring a supercharged Assistant powered by the latest large language model (LLM) technology and has planned some organizational changes to achieve this.
It appears that Google realizes it needs to catch up with other companies that have already demonstrated the capabilities of LLMs in their products. While LLMs have been used in chatbots and assistants, their practical evolution in this domain is yet to be fully realized. The current generation of voice assistants like Google Assistant, Alexa, and Siri function more like interactive Mad Libs, where users provide specific commands or questions, and the assistants respond accordingly.
The question arises whether an AI that can reference the entire Western canon for a simple query like driving time to the beach is a true improvement. While it may initially be entertaining to request weather updates in sonnet form, the novelty eventually wears off. Similarly, engaging in a conversation with a navigation system or discussing nuanced topics when seeking information on sushi joints might not be the ideal user experience for most people.
Instead, having an interface that can handle both straightforward commands and more complex interactions on-demand seems more practical. Google seems to be preparing itself for a future where its Assistant can adapt to user needs, whether they require a simple response or a more conversational AI experience. The pivot towards generative AI could bring about exciting developments in voice assistants and improve user interactions in the future.
