So apparently we have a prototype of a Virtual Assistant (VA) for our smart phones that can book us a table at a restaurant or help us finding a flight. You just have to ask Facebook’s M and he, she, it will do the thing. M will live in Facebook’s Messenger app and we’ll be able to ask it things by typing our questions. In theory, it is very different from Siri and such, because, as we can read in the blog post:
“M can actually complete tasks on your behalf. It can purchase items, get gifts delivered to your loved ones, book restaurants, travel arrangements, appointments and way more.”
Well, it was about time.
Berners-Lee wrote The Semantic Web almost fifteen years ago, describing a machine capable to help Lucy and Pete booking a series of physical therapy sessions. The Semantic Web agent would look through their agendas, best doctors based by rating and distance from respective homes and so forth.
In 2001 this seemed a BLAST, the future: something out there that we can only imagine and not grab. And now here we are, with M finding a table for us in the best restaurant of the city, with just a thirty-minutes query and the help of physical human beings.
I know it sounds sarcastic and that M is still a prototype, but this VA prototype brings to me at least three questions: What? How much? Do we need it?
What & How much
Berners-Lee’s vision of the Semantic Web was based on machines that could comprehend what we were asking and would be able to answer, basing their answers on data. Now, M can – or will – comprehend our questions and it’s able – or will be able – to give answers based on data.
The key here is the difference of the word “data” in the two cases. In Berners-Lee’s – utopian? – idea, data was provided by the people. The duty of the people was to encode their data, tagging them in a specific way, to be “understood” and be part of the Semantic Web. It’s the community that helps and works to improve itself, to improve the general knowledge of the world. Yes, like Wikipedia – that was actually theorized by Douglas Engelbart in the seventies.
I imagine M will rely on “public” and “secure” data, such Google Maps and I guess websites of museums and public places. This for the public. What’s for private? You know, Facebook is a business and its duty is to make business. Berners-Lee created the Web and it’s not an inch as rich as Zuckemberg – he’s not rich at all actually. It’s clear that Facebook is not spending its effort on M for the sake of the community so it’s planning to monetize. And I believe that if you run a private business – maybe that famous restaurant that M will book in thirty minutes – you will have to pay to be considered in M’s choices.
This is my concern about “what” data will be considered by a virtual assistant run by a private company who wants – rightly, don’t get me wrong here – to monetize. And of course the “how much” will it cost for privates comes along.
Do we need it?
Last summer I visited a small city by the sea with some friends. At dinnertime we wanted OF COURSE to eat some fish. So, what’s the best place for fish around here? Let’s ask TripAdvisor, suggested my three friends and dove in their smart phones. Yep, why not, I thought at first. Since I didn’t have the will not to even take out my phone from my pocket – summer, you know – I let them do the thing. But because of some connection problems, or alcohol problems – summer, you know – this stuff was taking ages. While wandering around I met my personal TripAdvisors: an old couple clearly not there for a spot vacation like us. They looked reliable and wise. So I just did the simplest thing in the world: I asked them. They ended up walking us at the place, and yes, the fish was excellent.
This simple story makes me wonder and wonder. The Semantic Web, the theory of Berners-Lee was focused on a simple thing: solve a problem in the simplest possible way. Scheduling a series of medical appointments is a problem, especially if you work and you have a tight agenda. Book a flight with the help of M – or any “virtual assistant” – will be solving a problem in an elegant way.
Is finding a restaurant a problem?
It can be in several occasions, but is it a problem when you’re on holiday, you have all the time in the world and you could just walk or ask an old couple and embrace this thing called serendipity?
Google Photos now “automatically organizes your memories by the people, places, and things that matter. You don’t have to tag or label any of them, and you don’t need to laboriously create albums”, as written on Google’s official blog. And people seem thrilled by this. So they can just upload the picture of their holidays and let someone else do the job.
Question is: Why the hell did you take so many pictures if you didn’t want them? Where is the pleasure if you let an algorithm choose for you?
I think that we’re walking on a dangerous path. Virtual assistants, Semantic Web, things like that, can greatly help us. They can really avoid us the tedious work and help us solving problems.
There’s a big but on the definition of a “problem” and the “borders” here. The border is passed when a “problem” is solved by a decision not taken by me, based on standard – or, better, standardized – data. That is the border I would never want to see myself passing: coping with a problem – that is not a problem – and do something just because the Virtual Assistant advised me to. It’s like reading a book because: everybody read that. In this case, “everybody” would be the “opinion” of our virtual assistant. When – not if, but when – VA will become sophisticated enough to be part of our daily life, we will have to remember that they are a tool, a way to solve our problems or to discover something new. A way, not the way.
Maybe it’s just me, but it feels important to mark this fact.
If we will rely more and more on Virtual Assistants we will risk giving up our wit, our creativity and, well, us. We will risk to standardize our parameters – aesthetic parameters, judgment parameters, entertainment parameters and so forth – like the standardized data on which VA’s answers are based upon. We will risk to end up becoming data collectors, letting algorithms do what we once believed it was the cool thing, like creating our own photo albums.