Facebook (META -0.09%) is killing M, its AI-powered personal assistant for Messenger, just two-and-a-half years after its launch. Facebook initially launched M in a closed beta in the San Francisco Bay Area in the hopes that it would evolve into a more sophisticated personal assistant like Apple's (AAPL -1.92%) Siri, Amazon's (AMZN -0.16%) Alexa, or Alphabet's (GOOG -0.21%) (GOOGL -0.30%) Google Assistant.
However, M's concierge services merely prodded other people to make calls for you, and didn't gain much traction among its beta group. In a statement to Engadget, Facebook stated that it "learned a lot" about what users wanted in a personal assistant, and would take "these useful insights to power other AI projects."
Image source: Getty Images.
However, M's premature death actually tells us a lot about Facebook's ambitions and shortcomings. Let's take a look at why Facebook developed M, and what its failure means for the company's future.
Why Facebook launched M
Facebook's biggest weakness is that it doesn't own an operating system like Apple's iOS or Google's Android. This makes it harder to expand its ecosystem, and cuts it off from launching app stores or monitoring a user's actions outside of the Facebook app.
Facebook tried to address this weakness many times, with HTC's ill-fated Facebook Phone, its Facebook Home launcher for Android devices, single sign-ons for apps and websites, Oculus Home, and the expansion of its Messenger app as a computing platform.
Out of these efforts, the Messenger app -- which reaches 1.2 billion monthly active users (MAUs) -- represents Facebook's best chance at establishing a "mini-ecosystem" within iOS and Android. Taking a cue from Tencent's WeChat, Facebook added stickers, mini-apps, mobile payments, and ride hailing services to the app, allowing it to partly bypass Apple and Google's mobile operating systems.
Facebook's Messenger app. Image source: Google Play.
Therefore, adding a Siri-like assistant to Messenger seemed like a smart way to draw data from Facebook users' profiles, gain additional information from their queries, and increase their overall dependence on the app. In theory, M would become smarter, and businesses would strengthen their ties to the platform -- which would reduce their dependence on traditional mobile apps.
Facebook clearly had the resources to accomplish this. It controls the world's biggest social network, and it launched its dedicated FAIR (Facebook Artificial Intelligence Research) unit in 2013 to upgrade the platform's AI capabilities.
Why M flopped
However, M flopped for two simple reasons -- it was late to the party, and its chatbot tech wasn't that impressive. Apple introduced Siri in late 2011, Google launched Google Now in 2012 (which became Google Assistant in 2016), and Amazon launched Alexa alongside its Echo smart speaker in 2014.
Facebook's introduction of M in 2015 came at a time when mobile users, who were split between iOS and Android, had grown accustomed to their default virtual assistants. Meanwhile, Alexa became a convenient choice for regular Amazon shoppers and Prime members. There wasn't much room for M to grow between those three giant ecosystems.
Things might have gone differently if Facebook's M and other chatbots worked. But its chatbots struggled to understand user queries, causing Facebook to scale back its Messenger chatbots efforts in early 2017 after they failed to fulfill 70% of users' requests.
The Information reported that those requests couldn't be handled without human agents due to shortcomings in the "technology to understand human requests." A major issue was that the chatbots were submitted by companies, and many low-quality ones were tethered to Messenger -- just as Apple and Google's app stores were once flooded with low-quality apps.
Is Facebook getting ahead of itself?
With 98% of its revenues coming from ad sales last quarter, Facebook is eager to diversify into new markets like virtual reality and AI chatbots. But the company is arguably getting ahead of itself in both markets -- its VR headset sales remain tepid, and its AI chatbots seem underdeveloped.
Looking ahead, Facebook probably hasn't given up on its efforts to create an "OS within an OS" with a virtual assistant. But it will probably need to take a different approach next time, since Siri, Google Assistant, and Alexa already dominate this growing market.
