I Blog About What I Know , What I Learn And What I Am Curious About.

Facebook Launches M a new virtual assistant.

Facebook unveiled its M personal assistant yesterday, shelled out a few details and revealed that it was currently being beta-tested by a couple hundred random users in the Bay Area. The service is unique in that it is partially human-powered, and these “M trainers” can complete real tasks like buying items and making reservations. Whatever M eventually turns out to be, it is clear that it is going to be different, and with so many buggy, mediocre digital assistants out there, different has to be a good thing, right?

Digital assistants as a whole seem too forced on consumers right now, and even M is a bit out of place among them. When flip phones introduced voice control a few years after the dawn of this millennium, people were impressed, but overall it was just a great nerdy dad gimmick to show off. Now, as tech improves, giant companies are treating this very seriously. M bucks the convention of being voice-activated, it lives inside the Messenger app and is primarily text-based. This is a mistake.
Also Siri, Google Now, Cortana and (kind of) Alexa all have moments where they’re uniquely helpful, but where they are similar is that they each (mostly) live natively on their own hardware devices.
Facebook’s M doesn’t have a physical home and unless Facebook’s planning on dropping a social OS anytime soon, it’s not going to. Digital assistants desperately need deep hardware integration in order to be useful because it takes too many steps to reach the functionality otherwise and digital assistants are tools that need to be accessible at a moment’s notice.

Feeling Lucky?

Digital assistants are essentially more personable, smarter search engines. In a lot of ways the only difference is branding and more transparency in exposing how much your digital assistant actually knows about you.

If digital assistants are search engines, than M is an “I’m feeling lucky” Google search where you’re probably just getting the one answer M believes is most likely to satisfy your need.
Facebook is already looking pretty naive with the examples they’re providing for how human interactions with M are going to look.

One person chats with M in a screenshot, “I’m going to Chicago next week. I’m looking for a great burger. Where should I go?”

In situations where you’re looking for an answer easily available in Google, like trying to find a great-rated burger in Chicago, I really really can’t understand why anyone would use M.

A) You have to type coherent sentences into M for the sake of a human understanding it, whereas in Google you could just type keywords, i.e. “best burger chicago.”

B) Search results are instantaneous, if it’s a question an M assistant has to search for, you’re literally waiting on them to Google something for you that you could just have easily done yourself.

The Human Element

The thing is: that’s a person on hold on your behalf. Facebook’s M trainers have customer service backgrounds. They make the trickier judgment calls, and perform other tasks that software can’t. If you ask M to plan a birthday dinner for your friend, the software might book the Uber and the restaurant, but a person might surprise your friend at the end of the night by sending over birthday cupcakes from her favorite bakery. “M learns from human behaviors,” says Marcus.
Eventually, the service might be sophisticated enough to figure this out on its own, but not soon. Right now, M trainers sit close to the engineering team inside Facebook offices. The company confirms the trainers are contractors but won’t say how many there are. Marcus anticipates that over time, Facebook will employ thousands of them, which will represent a substantial economic investment.
The company anticipates the cost will be offset by the revenue growth it is able to realize by capitalizing on M’s interactions. As WIRED’s Cade Metz explains, Facebook plans to use data generated by the service to feed much more complex AI systems that can reduce the burden on the trainers.

Facebook Call Center

Then again, would a human back end be such a bad thing? I recently spent hours (and hours) on hold with my bank after I had my card blocked for fraud prevention. It was utterly infuriating.
Marcus’ vision, as shared in this Wired article about M, is that I’d be able to ask M to sort it all out for me. Rather than having to sit on hold at 4 am myself (time zones, who needs them), an “M trainer” will do it for me.
He told Wired: “Facebook’s M trainers have customer service backgrounds. They make the trickier judgment calls, and perform other tasks that software can’t.”
He outlines a clear equation. The smarter the artificial intelligence, the less burden on the M trainers.

The more the AI can do, the less employees Facebook will need.
But they do seem to be interested in making M have a human touch.
Which at a time when the world is justifiably concerned about Silicon Valley firms inventing technologies that make entry-level or low-skilled jobs evaporate, the possibility of a global call centre is an interesting change of direction.
And for everyone else, it could be a system that could spell the end of hours on hold, endless security questions and voice-activated phone menus.
The downside, of course, is that you’ll be sharing more data with Facebook than ever before.
As we’re often reminded, losing elements of our privacy is the trade-off for using a service that’s free.
But how about this one: would you trade more of your personal data in return for a life without all the annoyances of customer service?

Leave a Reply

%d bloggers like this: