Facebook has officially announced a new foray into the world of AI-based, machine learning-powered personal assistants, this time in the form of a fully automated version of their M AI bot that can figure out what's going on in a message thread on Facebook Messenger and interject useful suggestions. M first debuted over a year ago, with many a human behind the scenes, fulfilling requests and feeding the system data. Thanks to the vast amount of data it accumulated from Facebook staff and users over the past year, M is ready to go fully automatic, and Suggestions From M is its very first adventure in doing so.
Suggestions from M works in a deceptively simple and intuitive manner. Users can't communicate directly with the system that powers Suggestions From M; instead, it listens in on conversations over Facebook Messenger, so long as a user hasn't opted out, and suggests actions or things to say based on cues it recognizes in the conversation. A friend asking for money may prompt M to suggest sending the amount they asked for, or saying something like "Let's have dinner Friday night" may prompt M to suggest creating a plan for said dinner. M can also suggest emoji and stickers during a conversation.
Suggestions From M can be followed with a tap, bringing you to the specified action, or dismissed with a swipe. Alternatively, users can long-press on a suggestion to see more options related to it. There is also an M Settings menu in the Messenger app, where users can easily turn M off altogether. The feature is currently rolling out gradually across the United States over an unspecified period, and once it's rolled out to all users in the US, it will likely roll out to other territories in due course. For now, there appears to be no way to force Suggestions From M to come to your device, so those who are interested and don't have it yet, will simply have to wait until Facebook gets it out to them. According to Facebook's press release about Suggestions From M, the AI assistant is going to be getting more new functions in the future as it learns more, is developed further, and has enough data to cover a wider variety of situations and functions.