Rare is the person who currently doesn’t ask ChatGPT what to write, how to respond to an email, or how to explain an idea they have in their head. It is undeniable to say that this is convenient, fast and works. But be careful.
Ramón López de Mántaras, one of the great pioneers of artificial intelligence in Spain, warns: there is a basic ability that human beings, unwittingly, are putting at risk.
on the podcast Lives of Othersthe expert wanted to focus on something very specific. It is about the indiscriminate use of language models like ChatGPT and how this can directly affect cognitive abilities.
It’s not that he believes that technology is bad in itself, The serious problem behind all this is that humanity is beginning to use it for things that should not be delegated to AI..
As he exemplifies, teaching a child to add using only a calculator may give him the correct result, but it does not teach him to understand what he is doing. You get what you want, but learning doesn’t happen.
«Writing is thinking; writing is not vomiting one word after another, like ChatGPT does,» he comments.
«If we delegate writing to Artificial Intelligencewe not only delegate the writing, we delegate thinking. We can lose the cognitive ability to think. Be careful when using these language models happily.»he warns.
His idea is clear: if you let an AI write for you, you are not just delegating a mechanical task. You are leaving aside your own reflection process. According to López de Mántaras, that is the real danger, getting used to not thinking because someone or something does it for you.
«Consciousness and intelligence can only occur in living beings»
For the expert, Consciousness and intelligence, in the full sense, continue to be the exclusive heritage of living beings..
«Scientifically speaking, I see rather little interest in it. If we stick to modern AI, the big scientific question is whether it is possible for there to be intelligence in something that is not biological. I believe that, to really understand the world, it is necessary to have consciousness,» he comments.
«My opinion, and that of others like Yann LeCun or Demis Hassabis, is that generative AI not only does not move us towards that strong AI, but, as LeCun says, it diverts us from the path to achieving it. The argument is that they cannot understand the world because they do not have a body with which to interact in this world,» he adds.
According to López de Mántaras, the big question remains unanswered: can there be intelligence in something that is not biological? For him, the key is consciousness, that ability to understand the world and have subjective experiences.
The argument is that machines, no matter how advanced, have no body or sensory experiences. Without that physical interaction, they cannot learn the basic laws of the world or develop real common sense.
«Having a body allows us to learn elementary cause-effect relationships, like if you let go of an apple, it falls to the ground. But that does not imply consciousness,» he points out.
For those who defend that AI is already conscious or intelligent like a human being, the expert is blunt: «Whoever claims that the machine is intelligent and conscious, that is the one who has to prove it. I am one of those who believe that being conscious, understanding the world and being intelligent in the human sense of the term is deeply rooted in the fact that we are living beings.»
To all this we must add another of its great concerns: regulating AI. For the researcher, the real risks of AI are not that it dominates humanity, but rather the manipulation of information, misinformation, biases, discrimination and unsustainable energy consumption.