The appearance of automatic content generators through artificial intelligence such as Wall-E and GPT have caused many to speak wonders of what artificial intelligence is. Some even speak of the end of certain trades, some in a positive way and others in a negative way. However, for us the crux of the matter and where the basic problem lies.
Why is AI being miseducated?
The discipline within the world of artificial intelligence that is growing the most is that of the so-called Deep Learning or Deep Learning, this is based on giving the system a large amount of data in one way or another so that the system ends up detecting a pattern. However, it does not have the capacity to learn with a little information and needs a large number of samples.
For example, suppose you have a cute pussycat and you show it to a very young boy or girl who has never seen one before. This first experience will serve the creature to store in its memory what a cat is and its characteristics, and it will be able to identify it in other media, such as in the form of a drawing, a sculpture. All this due to the fact that our brain has that capacity, but unfortunately, in the case of computers, not even the most powerful in the world has that capacity.
In other words, we are going to need a huge number of photos of cats for an AI to identify what a cat really is through Deep Learning. Of course, our example is completely innocent and innocuous. However, a large amount of information requires that it be cured, that is, selected so that the conclusions or patterns that the system concludes are not incorrect. That is, just as we do not teach a child certain things, we are also interested in not being rude to the AI so that it does not generate the wrong patterns.
Internet is a garbage pit
In order to train various artificial intelligences, especially conversational ones, many of their creators, in order to save time and money, decide not to train them with scientifically verified data, but instead leave them in the hands of Internet users with more than disastrous consequences. The reason? You can imagine it, from people interacting with false and unverified information, to people wanting the system to have negative thoughts about issues such as gender identity, sexual orientation and also about other ethnicities that end up being problematic as they are sources of information from hatred.
The fact of not curating the information is sensitive, for example, we can teach an AI to draw the human body with several different samples. However, we all know that the use of nudity is necessary. And where is there more nude material? Well yes, you guessed it, in the pornographic industry. So there is a danger that an AI trained with these images ends up generating one that is very similar to the source material or highly influenced by it.
That is why it is important not to give AI any criteria of authority and to ensure that they work with the correct information at all times. Many will see it as censorship, but it is the same process as when we educate a minor. That is, the internet is a garbage pit that is not suitable as a database for artificial intelligence. So it is important to ensure the verification of the data that we use as learning patterns.
No, artificial intelligence is not going to take your job
In recent days, we have seen that they do not stop repeating the mantra that, for the creators of content, whether written or in the form of images, the world is ending due to the fact that artificial intelligence will end up replacing our work. What they are unaware of is that these machines have the inability to create and can only act as imitation elements from what has already been previously generated. Namely, they do not have any creative ability, which greatly limits their ability.
We must start from the fact that we are still very green when it comes to artificial intelligence, and much of what we are seeing announced in recent years, despite the fact that it may seem impressive to us, is in very early stages and we do not have They have to treat it as reliable information, but as conclusions of an AI that draws partial conclusions from a very limited and sometimes even erroneous spectrum of data.
Many will affirm that curating the data and selecting it is going against “freedom of expression”, but it is one thing to give an opinion on true information and another to draw conclusions from falsehoods or a partial vision of reality. In these times it is important to know the limitations of the technologies we use and it is especially important not to treat artificial intelligence as if it were some kind of oracle or reader of reality.
The consequences of being rude to AI are dire
To many it may sound like a joke, but imagine that we train an artificial intelligence that serves as a first aid assistant. That is, what to do in the event that a person is injured. If the data chosen is wrong and we find that we are rude to an AI, then it can give advice that could have dire consequences for both the injured and the helpers.
And it is not the only case, the fears and fears towards certain minorities can be passed off as information and freedom of expression and this can also have a cost in the form of human lives if there is no control. Let’s not forget the continuous murders and exterminations. So it seems to us a huge mistake to be rude to AI with the internet. In the end we will end up paying for it, and very expensive, given that we are giving authority to a system that only knows how to find repetitive patterns and that is no different in terms of conscious knowledge from a parrot, what is more, it is even less intelligent.