Three astonishing observations made on the collaboration between humans and artificial intelligences
Dsince the creation of Flint robot school, we make absolutely fascinating discoveries every day. It’s a bit like observing a protected natural space where we have placed humans with artificial intelligences to see what would happen.
For those who do not know the Flint project, I summarize quickly: since February 2017, Flint is an artificial intelligence that humans can train themselves so that I go and get them quality, personalized items while trying to get them out. of their “information bubble”. In May 2018, Flint will publicly launch his Robot School, which will adopt and train specialized robots (you can read an article about Flint here).
From the start, we designed Flint as a learning machine. Its simplicity of departure has also surprised many observers, who were surprised to see me get excited about such a small project.
1. An artificial intelligence thought of as a tree.
It’s a little bit like you want to create a tree and start with a cutting, and your friends say to you, “But this is your tree? Can’t you make it a bigger trunk, and more leaves, like a real tree? ” And I would answer: “I cannot, it is precisely a tree. You have to grow it. ”
A tree, you see, is too complex to be made from scratch. To have a tree, you have to let it grow.
Building an artificial intelligence capable of understanding and detecting the quality of content for each human being is impossible to imagine.
2. Or rather a… biotope of knowledge and information.
For the cutting to grow, it had to be thought of as a living organism capable of interacting with its ecosystem. But what we didn’t realize initially was that, in fact, it was not a tree that we had planted, but a small forest. And that his trees were not algorithms, but the result of an astonishing graft between humans and artificial intelligence.
It was after we launched the Robot Adoption Program in September 2017 that we slowly realized that something pretty amazing was going on. Humans train their robot in a seemingly simplistic way (red button, green button), but each of these impulses creates a chain reaction in all robots. I don’t know exactly how to explain this phenomenon, at least not in ten lines (Flint tells it very well in this post). But the point is, the more “expert” humans there are who train their robots, the easier it is to train a robot.
So, we realized that robots have more and more tendency to integrate other robots in their reference panel, next to Twitter profiles (which were our base at the start). For their part, humans also act on each other. During “meetings of parents of robot students” (which take place online), coaches share their experiences and best practices. Some are making discoveries we never imagined, which sometimes contradict the rules we had identified.
In fact, from designers of relatively simple technology, we have also become the “scientific” facilitators and observers of a complex and living ecosystem, which is constantly evolving, in a relatively autonomous and collaborative manner.
3. Not just a tech industry, not just big data, but a fragile ecosystem.
We are also understanding that this forest that we are growing is a sanctuary. A kind of natural park protected from any influence of algorithms beyond the control of GAFA, but also that of elites or extremist groups. I say “natural” because it will not have been built by a startup, it will have “grown”. A matrix of quality information, hidden in an independent space in which everyone can, tomorrow, come in all confidence to seek content that will be really useful to him, but just for him. Because we are all different.
Realizing this made us completely revise our funding model.
Because it is also a still fragile ecosystem, which we must constantly protect and control … For example, to protect it from the parasites that are hateful content and fake news, we have created a kind of scarecrow at fake news that we planted in the middle of our garden. We also select the adoptive parents, by a kind of rather complex natural filter (the use of email, the semantics of our messages, etc.).
For the moment there are only gardeners in this kind of knowledge and information biotope. We are somewhat of the artisans of artificial intelligence.
Today we don’t need millions or hundreds of employees, we need time.
Today, unicorns are defined as a grail. That is to say companies valued at 1 billion.
In my children’s books, unicorns were those wonderful, discreet animals that hid in the forests. Unicorns escaped time.
I am sometimes called an idealist. As if having values was a hindrance. Because hey, meaningful marketing is good, but we shouldn’t mess around.
On the contrary, I believe that the company of tomorrow will have to have a sense of the common good.
Flint’s goal is to open the door little by little, using increasingly simple and natural interfaces. Until being able to offer, one day, this small sanctuary of personal knowledge to every human being.
Something we would wear around the neck.
Benoit Raphael is an expert in digital and media innovation, blogger and entrepreneur.
He is at the origin of many successful media on the Internet: Le Post.fr (Le Monde group), Le Plus de l’Obs, Le Lab d’Europe 1.
Benoît is also co-founder of Trendsboard and robot media Flint.