It really is time we started to understand AI, like now

The main problem with artificial intelligence (AI) is that the vast majority of those who talk about it haven’t the faintest idea what they’re talking about. That’s the largely correct premise of a good article in Futurism entitled “You have no idea what artificial intelligence really does”, which I quickly empathized with because it begins by talking about Sophia, the robot created by Hanson Robotics I had the opportunity to interact with earlier this year, and which was what I expected: a fantastic interface, brilliant communication, but nothing to do with a supposed robot that supposedly processed and understood conversations, when all it was doing was reading scripts managed by a person via a laptop.

According to recent surveys, 42% of managers believe that IA will be critical for their companies in the next two years. The first thing you feel when you see such a number is pity and commiseration for the 58% percent of the remainder who don’t, because they will very likely lose their job or suffer the consequences of ignoring a revolution. At the same time, many of those who believe that AI will be important do so on the basis of hype, because of the thousands of articles that talk about the subject — along the lines of “there must be something to this” or “my competitor has achieved this or that through AI”, but without really knowing what the hell they are talking about.

Part of the responsibility is the word itself, artificial intelligence: a vague, very broad term that includes a wide variety of technologies that, when they reach a minimum degree of development, establish themselves as their own disciplines. In practice, we end up calling AI practically anything, and in addition, we illustrate it with a photograph of Terminator and go off on a tangent about how we will soon be surrounded by robots with human capabilities hell bent on destroying the human race.

What should a manager know about AI or its components to be able to assess its importance? Basically, we are talking about technologies capable of taking data and, after transforming it properly — a task that should not be underestimated and that, in fact, takes a huge part of the effort and the resources invested in any project of this type — is capable of classifying that data, assigning it with a series of labels, and processing it in a way that is capable of deriving rules, some of which can be very interesting. What does “your next appointment with the doctor could be with artificial intelligence” mean? Quite simply, that we already have reasonably mature technologies capable of taking a series of diagnostic elements — many of which I am convinced will come from simple wearables that many of us are already wearing — and use them to highlight certain possibilities. Algorithms are already more reliable than people at detecting tumors in medical images, as well as analyzing heart rate data or ultrasounds, in addition, they do so by analyzing these data constantly, tirelessly, without distraction, far beyond even the most experienced and expert human. In other words, robots aren’t going to be sitting behind the desk at a clinic. The first are real and existing technologies, the second is cheap sensationalism. But it is still important, and above all, potentially important in defining many of the scenarios we will see in the future.

Can Atlas, the Boston Dynamics robot, do Parkour? Yes, it can, demonstrating an agility and ability I lack. But from there to processing the information around it along the lines of robots in the movies, is a stretch. Each new capacity added to Atlas is a new development, complex and unrelated to the previous one. The idea of creatures equipped with something minimally similar to human intelligence is so far away as to be pure speculation.

But we are already able to use a technology to analyze larger and larger amounts of data and after a complex transformation process, we can see indicators of possible anomalous patterns, the potential value of certain clients, we can analyze important financial patterns, create analytics of certain events over time, produce image recognition, detect events that require maintenance, and generally make many kids of prediction, with its limitations, like all technology, but also with its potential. If you want to understand what deep learning involves, for example, check out this article. If you can see the importance of unsupervised learning or distinguish its different uses, read this one. There are many more out there, and while they are no substitute for solid training, they will at least give you an idea of what we are talking about, rather than seeing the issue in terms of Terminator stealing your job. AI has incredible potential and will change the way we work and live. AI will likely change the world as much, if not more, than the internet has: we need to start seeing its potential, now; we need to join those who can see, and not those in the movie The Others: “They’re dead, but they don’t know it.”

(En espa?ol, aquí)

?