Artificial intelligence: nothing to do

Inanimate objects are hell-bent on outdoing us at anything we think we do better than them. Its arithmetic speed leaves us at the level of bitumen since the time of the slide rule, that wooden instrument with which the director of theoretical physics at Los Alamos, Hans Bethe, solved the equations to manufacture the first atomic bomb. My father used it, but my generation was one of the first to switch to the pocket calculator. Then they started beating us at chess, checkers, Go and even poker, which is already going crazy. As any gambler knows, winning at poker doesn't depend as much on the cards you're carrying as it does on what your opponent thinks you're carrying, and this high-level malice was considered the exclusive privilege of humans until a few years ago. But it's not like that.

Have you noticed that no one complains about someone's robotic behavior anymore? The phrase was valid when the machines spoke like R2-D2, the robotic block from

, but it is evident that George Lucas fell grossly short in that linguistic section of the series. Today's robots talk so well that he even worries they might pose as a prime minister or a brigadier general to sow chaos and confusion. Competition from large language models

LLMs, like the one underpinned by ChatGPT) was the main reason Hollywood screenwriters went on strike last summer, and restrictions on their use were the key to They called off the protests. The video generation systems that the technology companies OpenAI and Google have developed suggest a panorama in which not only the scriptwriters, but the filmmakers and editors themselves begin to fear for their work, as translators and accountants, architects and stockbrokers, poets and lovers.

Taking pessimism to the extreme is wrong, however, as well as useless. Having a machine beat us at something is not the end of the world. The love of chess, for example, has only grown since

beat Kasparov in 1997. In fact, the vast majority of people did not need

to lose at chess — it would be enough for me to play against a Rhesus macaque—and the cerebral tribulations of a world champion don't matter enough to us. The fact that a machine can beat you should no more deter you from playing chess than the mere existence of Carlos Alcaraz deters you from playing tennis. Routine jobs are another story, as many people would prefer not to do them, like

. Working in a ditch or repairing the subway at four in the morning shouldn't be a chore either, and LLMs have very little to say there.

When a topic has a lot to talk about, read everything that has to be said.

Subscribe here

But that's not the new thing we've been talking about for a couple of years. Low-skilled jobs, in fact, were the first to be transformed by robotics, especially on the assembly lines of the automobile industry. Jobs were lost there and others were created with little investment in continuing training. Now we talk about doctors, surgeons and pharmacologists, film directors and screenwriters, painters, thinkers, creators and everything greatest that our species can exhibit before cosmic history.

But let's end with some good news. ChatGPT and other systems of this type are not going to grow exponentially. These models are improving so far by swallowing texts from the Internet (more texts, better results), but they have already swallowed almost everything. A stalemate is approaching.

_