Facebook’s Chief AI Scientist Says the Service Would Like to Offer Smart Digital Assistants. Here’s Why.
Intel made news at CES when it revealed that it was working with Facebook on an AI chip. Yann LeCun, Facebook’s chief AI scientist, “indicated that [the company] was developing its own custom ‘ASIC’ chips to support its AI programs” at a conference in San Fransico yesterday according to Financial Times. This effort is in addition to the work with Intel and is driven by the need for faster processing power.
In an interview with Financial Times, Mr. LeCun suggested that AI innovations in the past were often driven by hardware advances and Facebook was focused on “anything we can do to lower the power consumption [and] improve the latency.” This could be applied to analyzing the 2-3 billion photos uploaded on Facebook daily, something that is already done to identify user faces for tagging and to flag images that contain nudity. However, LeCun didn’t stop at suggesting existing use cases could be improved. He commented:
In terms of new uses, one thing Facebook would be interested in is offering smart digital assistants — something that has a level of common sense. They have background knowledge and you can have a discussion with them on any topic.
FACEBOOK DOESN’T WANT TO SURRENDER IN THE AI ASSISTANT RACE
Facebook has long had ambitions to launch its own AI-based assistant. Company CEO Mark Zuckerberg famously built his own assistant in 2016 that was loosely based on the Jarvis from the Iron Man comic and movie series. Facebook even had is own assistant, M, which it shut down in January 2018. M was human-powered with some automation that Facebook said in 2015 would expand beyond beta users and become more automated over time. However, it didn’t expand and Zuckerberg eventually told The Verge that it was a useful “experiment” and they “learned a lot.”
When Facebook introduced the Portal smart display, it did include some basic voice recognition features proprietary to the company. It also included Amazon Alexa for answering general questions, accessing music, and controlling smart home devices among other features. By “learned a lot” from the M experiment, it may be that Facebook tempered its ambitions about becoming a general purpose voice assistant that covers a broad set of consumer use cases. Instead, Portal and Facebook’s future assistant may instead focus on niche domains closer to its core feature set that is useful and differentiated.