Human beings, or being human

Some days seem to conspire to tell you a story from several points of view.

The story began in the car on the way to Milan, heading to a leadership course, while I was listening to a podcast about AI. Two ideas reached me: the first, that the more we learn to talk to machines, the clearer it becomes how little we know about talking to each other; the second, about how useful or advisable it is to humanise machines.

Shortly afterwards, at the Otto Walter course led by the excellent Matteo Corbetta, came an expression that stuck with me and that I report as I remember it: the “happiness formula” as a guide to building motivation in a company. Here came the first link back to the drive along the “Milano-Laghi”: almost every variable in that formula has to do with the way we talk to each other, with what we say and how we express ourselves. How much we are recognised and how much we take an interest in others necessarily passes through the questions that are asked and the ability to truly listen. This point of view convinced me so completely, in its unmistakable truth, that I suggested, partly as a provocation, that happiness should become our first indicator: not the last item anyone remembers at the end of the year, but the first KPI.

The second signal comes from a reflection by Nello Cristianini on how people try to make a machine “good”. The gist, as I understood it and as I summarise it here, was that the soundest way to keep an artificial intelligence from harming us is not to fence it in with a list of prohibitions, but to instil in it something resembling an ethic, almost a conscience, because that reaches deeper than any rule. It is the same principle that leads those who train these systems to ask them not only to obey, but to care about the wellbeing of whoever is in front of them. And there a thought became clear to me: training a machine on a feeling as deeply human as caring is perhaps the most powerful way we have to make it reliable. More powerful than any instruction, because an instruction is simply carried out, whereas care, when it is truly rooted, guides everything else.

Here comes the other crossing: at the course we work on hypothetical appraisal forms, performance reviews and related topics. Nothing particularly complex, when the discussion is about objective, performance-related matters, in a word about hard skills. But when the topics concern behaviour, character, attitude, the personal sphere? Well, that is where empathy comes into play, human interest, goodwill… provided that, to borrow an image I remember from the course, the “Gorilla” instinct inside us does not take over.

The moral? We work to teach machines to care, we consider it the deepest thing we can imprint on a system, and meanwhile inside our organisations “caring” is often the first item we cut. We measure everything and ask little. We fill our days with dashboards and empty our meetings of listening. We are imprinting on silicon exactly what we are letting die out in ourselves.

And yet this is precisely where I see the opportunity, more than the threat. We can hand over more and more of the hard part of the job to the machine: calculation, repetition, the processing of mountains of data. What remains to us, and at last with more time to cultivate it, is the soft part, which is soft in name only. On one side, intuition, that is the ability to read in the numbers what the numbers alone will never say. On the other, sensitivity: empathy, attention, caring in our relationships with those other selves around us. It is the part the machine is laboriously trying to learn from us, and neglecting it now, just as we discover how much it weighs, would be the real mistake.

Perhaps the formula we were looking for was all here. We can remain human beings and forget, little by little, to be human: cultivating the soft part is the way never to confuse the two.