Facebook Developing Virtual Assistants to Summarize News, Read Minds: Report
Facebook is building a virtual assistant to digest, summarize, and read articles for users, according to a Buzzfeed report on a closed company meeting. The TL;DR tool, internet slang for “too long, didn’t read,” would use AI to condense articles into bullet points and read them out loud for users who want to skip reading it for themselves. The social media giant also discussed other planned AI projects, including creating a neural sensor to detect thoughts as they form and turn them into commands to AI assistants.
The product manager at the Tuesday meeting unveiling TL;DR described how it would apply Facebook’s AI algorithms to reading news articles to pinpoint the most crucial elements and recreate them in understandable bullet points. Facebook’s voice assistant could read out just the bullet points or the entire article. The AI will supposedly also answer questions using the information in the piece as a database. As an example, Facebook described someone looking at an article about a COVID-19 vaccine and asking how many people it would help. The AI would then respond with just the number mentioned in the article.
ah the fourth horseman of the apocalypse is here,, https://t.co/KHpNtHtR3a
— Patricia Hernandez (@xpatriciah) December 16, 2020
i don’t know about you guys but i for one have always trusted the information i’ve received from facebook https://t.co/dNrYuDBjmz
— Edward Ongweso Jr (@bigblackjacobin) December 16, 2020
The AI: “I have learned every fact in existence and can solve any problem you put to me. What is thy bidding?”
Humans: “Will you read GQ for us?” https://t.co/AnVujVk6Hm
— Eric Lach (@ericlach) December 16, 2020
The Twitter responses from reporters and others in the news media were mostly critical and sarcastic, as the small sample above indicates. That’s something the AI will need to consider if it is summarizing articles with more defined voices and tones that are not straightforward. Otherwise, it faces the same issue that prompted Members of Parliament in the United Kingdom to ask Amazon about Alexa being antisemitic. The voice assistant was documented citing conspiracy websites and reciting Wikipedia articles out of context in a way that seemed to endorse antisemitic stereotypes. It’s all too easy to imagine an AI summarizing an article about a fraught topic and accidentally taking an opinion or sarcastic comment as a reported fact. Throw in The Onion or humor columns and give Facebook every benefit of the doubt possible, but it’s likely there will be at least a few negative headlines about Facebook for TL;DR to summarize
Reading Minds and Translating Thoughts
Facebook is also working on skipping voice entirely and going straight into connecting people’s thoughts to virtual assistants and smart devices. The company shared an update on a neural sensor that can detect signals from the brain that connect to physical action. For instance, thinking about typing could make letters appear on a screen or move a video game character around.
The company’s leaders also talked about releasing a universal translator. That idea has been percolating for a while, with Facebook open-sourcing its language translation model that can shift between any two of 100 languages its automatic speech recognition model capable of understanding 51 languages, built on more than 16,000 hours of voice recordings. The goal is to make it possible for voice assistants to grasp both what someone is saying and what language they are speaking. Combined with 100 languages for translation, Facebook could get a step ahead of voice AI rivals like Amazon and Google. The translations might also be a boon for another open-source project, the Blender chatbot, which is supposed to be able to hold a conversation on any subject and show empathy with users. All of these upcoming tools may be part of the operating system Facebook has been quietly working on, and be a way to promote the voice assistant Facebook has confirmed it is developing.