(These are excerpts from my book "Intelligence is not Artificial")
Another Philosophical Intermezzo: Do We Really Want Intelligence at All?
Intelligence is messy. When we interact with human beings, we have to consider their state of mind besides our immediate goal. We may only need a simple favor, but it makes a huge difference whether our interlocutor is happy or sad, on vacation or asleep, has just lost a close relative or been injured in an accident, angry at us, busy with her work, etc. Whether our interlocutor is capable or not of performing that favor for us may be a secondary factor compared with whether she is in the mental condition of doing it and doing it right now. On the other hand, when we deal with a dumb machine, the only issue is whether the machine is capable of performing the task or not. If it is, and the power chord is plugged into the power outlet, it will. It won't complain that it's tired or in a bad mood, it won't ask us for a cigarette, it won't spend ten minutes gossiping about the neighbors, it won't comment on the government or the soccer game.
It may seem a paradox, but, as long as machines are dumb, they are easy and painless to interact with. They simply do what we ask them to do. No whims. No complaints. No formalities.
The complication that comes with intelligent beings is that they are subject to moods, feelings, opinions, intentions, motives, etc. There is a complicated cognitive apparatus at work that determines the unpredictable reaction of an intelligent being when you ask even the simplest of questions. If your wife is mad at you, even the simplest question "What time is it?" might not get an answer. On the other hand, if you use the right manners at the right time, a complete stranger may do something truly important for you. In many cases it is crucial to know how to motivate people. But in other cases that is not enough (if the person is in a bad mood for reasons that are totally independent of your will). Human beings are a mess. Dealing with them is a major project. And that's not to mention the fact that human beings sleep, get sick, go on vacation, and even take lunch breaks. In Western Europe they are often on strike.
Compare humans with dumb machines that simply do what you ask. For example, the automatic teller machine hands you money at any time of the day or night any day of the year. Wherever the intelligent being has been replaced by a dumb machine, the interaction is simpler. We structured the interaction so that the dumb machine can perform all the operations that we need.
The reason that automated customer support has replaced human beings in so many fields is not only that it is cheaper to operate by the provider but also that it is preferred in the majority of cases by the majority of customers. The honest truth is that very few of us enjoy waiting for an operator to tell us "Hello? How are you? Isn't it a beautiful day? How can i help you?" Most of us prefer to press digits on a phone keypad. The truth is that most customers are happy if we remove the complication of dealing with human beings.
When i worked in the corporate world, my top two frustrations were the secretary and the middle management. Dealing with the secretary (especially in unionized Italy) required superior psychological skills: say the wrong word in the wrong tone, and she'll boycott your entire day. Most middle managers were mediocre and mostly slowed down things, seemingly paid mainly to kill great ideas. The only way to get important things done quickly was, again, to use the art of psychology: befriend them, chat with them, find out what motivated them, offer them rides home, hang out with them. Don't i wish that my colleagues and my secretary had been heartless robots?
And, let's face it, we often don't have the patience for human interactions that involve protocols of behavior. We are often happy when good manners are replaced by cold mechanic interactions shaped by goals and restrained by laws. Hence we do not really want machines with human intelligence, i.e. we don't want them to have emotions, to be verbose, to deceive, to plead, etc. One goal of inventing machines is precisely to remove all of that, to remove that inefficient, annoying, time-consuming quality of "humanity".
We removed the human/intelligent element from many facets of ordinary life because the truth is that in most cases we don't want to deal with intelligent beings. We want to deal with very dumb machines that will perform a very simple action when we press a button.
I'll let psychologists and anthropologists study the reasons for this trend towards less and less humane interactions, but the point here is that intelligence comes at a high price: intelligence comes with feelings, opinions, habits, and a lot of other baggage. You can't have real intelligence without that baggage.
When we study how to create intelligent machines, do we really mean "intelligent" or do we mean "stupid in a way that will serve our intelligence"?
Back to the Table of Contents
Purchase "Intelligence is not Artificial"