Blog AI
Artificial Intelligence mistakes. Is AI in its infancy?

Artificial Intelligence mistakes. Is AI in its infancy?


In recent months we have watched with amazement, fascination and even some concern the spectacular progress of artificial intelligences.
You have already analyzed in depth many of their possibilities in the present and future, but now it is time to know the other side of the coin, that is, when these types of intelligences are wrong and do not really seem so intelligent.

Before we naively thought that only humans had the ability to create art, but artificial intelligence programs such as DALL-E, StableDiffusion and MidJourney have created works that have little to envy to those of the great masters of universal painting and also impressive images that at first glance look like real people and things.
Yes, it is undeniable that these tools have managed to convincingly represent human beings.

But since no one is perfect, AIs still fail in key aspects such as the number of teeth. Oh, and artificial intelligences also have a particular problem with the recreation of hands, which are often represented with too many or too few fingers or with really grotesque shapes.
The reason? As any painter or sculptor knows, human hands are very complex and require a deep knowledge of biomechanics and movement dynamics.

Still on the subject of images, apps that use artificial intelligence to turn real people into anime-style characters have become popular, with surprising results. But as we see in this amusing selection, the problem is that sometimes the AIs get confused and turn animals into humans or vice versa, among other hilarious mistakes.

And while we’re on the subject of funny mistakes, a couple of years ago, the Scottish second division soccer team Inverness Caledonian Thistle decided to replace their human cameramen with AI-operated automatic tracking cameras, which were always supposed to follow the ball, and, consequently, the actions of the game. Unfortunately, those watching this game from home missed much of what was important, because the AI mistook the linesman’s bald head for the ball, and mainly followed him.
It is said that the fans, apart from complaining, requested that this referee be provided with a wig.

In reality, the list of gross errors of various artificial intelligences is long.
And they have been making mistakes for years that many did not find so amusing, such as showing inappropriate content to children.

Rejecting passport photos of Asian people by considering that they had their eyes closed, or getting very, very racist, or even praising Hitler, as happened with a Microsoft chatbot, fed by inappropriate comments from Internet users.

And since a few months ago, the popular ChatGPT became the great hope of millions of students around the world, but in truth, one should not rely too much on the answers of this chatbot still under development.
For example, ChatGPT’s wrong answers to children’s riddles have caused more than one smile. On some occasions ChatGPT has been pulled when answering anything at all.

And if you want to check more conclusively the mistakes of this AI, just ask for a biography of me or any TV Host you admire or want.
I have done so on more than one occasion, and among countless false and incoherent data, clear proof that AIs still can not match our intellect, and although they have already become important tools in more than one field, they still have much to learn, because as we have just seen, to err is not only human.

Agregue un comentario

Su dirección de correo no se hará público. Los campos requeridos están marcados *