Artificial intelligence scientists look like teenagers – 09/04/2023 – João Pereira Coutinho

As a layman, I laughed out loud at the public letter that several sages signed calling for a pause in the development of artificial intelligence.

The topic is serious, it seems. But the attitude of the staff is reminiscent of those teenagers who, well drunk and well smoked, suddenly discover that things have gone too far. The parents are arriving and the house is on fire.

According to the wise, AI laboratories, in a hellish race, are producing increasingly powerful digital brains and, an important detail, whose operation nobody understands well, much less controls.

Because of this, they ask for a six-month “moratorium”, just to cure the hangover. Didn’t I say they look like teenagers?

And these are the optimists, mind you, who still believe in “moratoriums”. In Time magazine, computer scientist Eliezer Yudkowsky goes further and calls for an end to artificial intelligence research. Either that or we all die, he says.

Don’t ask me where the truth lives. As a Woody Allen character would say, I don’t even know how the can opener works.

On the one hand, the world has been fantasizing about its own extinction for too long. It could be a computer bug at the turn of the millennium. Or a pandemic. Or global warming. Or a nuclear war.

In that sense, adding artificial intelligence to the bill might sound like old-fashioned history.

On the other hand, I swallow hard when I see the almost centenary Henry Kissinger coming out of retirement to sign, along with Eric Schmidt and Daniel Huttenlocher, two experts on the subject, a book like “The Age of AI: And Our Human Future”.

Anyone interested in the subject should start here, not with the public letter: the tone is calm, pedagogical, moderately optimistic. And sensible.

The first wisdom lies in the fact that it is not possible to “uninvent” things. When the genie comes out of the bottle, it scatters around.

A six-month moratorium is of no use. Or until the end of artificial intelligence research by well-known labs. There will always be someone, somewhere, operating underground.

But the suspension or ban also does not work because artificial intelligence, which is already present in a thousand activities or industries, is an astonishing epistemological advance.

As defended by the trio of authors, the main advances that humanity has registered have always been an addition to something that already existed.

Films are moving photographs. Cars replaced carriages. Battle tanks are in the cavalry continuity. Even nuclear weapons, in essence, take the destructive capacity of conventional weapons to an extreme.

What seems to distinguish artificial intelligence is not just the ability to process obscene amounts of information. It is the fact of detecting aspects of reality that escaped human beings. A blessing?

It could be: in medicine and other natural sciences, the leap will be gigantic in the discovery of new techniques or therapies. It already happens.

The impact will be more problematic in areas where human oversight cannot be ruled out. Like war, of course, where self-doubt and “morality” can make the difference between survival and catastrophe. A digital brain without doubts (and without morals) is an invitation to hell.

As the authors defend, there are three types of relationship that we can establish with the new machines: we can limit them; act in partnership with them; or submit.

There is no ready answer for all scenarios. There will be times when the machine will have an obvious ascendancy over humans — in detecting disease, for example, and even in the pursuit of certain pharmacologies.

There will be other times when artificial intelligence will be more of an auxiliary instrument of human action, as has always happened with new technologies.

And there will be circumstances in which the last word should be ours, not some algorithm’s. Will we be able to establish these differences?

Mystery. But I suspect it would be more fruitful to discuss and apply the answers to this question than to fantasize about lost innocence.

PRESENT LINK: Did you like this text? Subscriber can release five free hits of any link per day. Just click the blue F below.