The artificial intelligence boom has caused a fascination around it, suddenly many see enormous potential applications for it, while others are terrified of it. Although we usually talk about the benefits of it from our articles, we cannot forget that not everything is good and that there is a certain level of danger in this new technology. But, Is AI a danger to people?
Recently in Italy the use of ChatGPT has been banned due to the panic that it supposes that it replaces certain jobs. However, these legal decisions are nothing more than the product of not understanding how a technology works and of completely exaggerating its capabilities, or rather that we live in times where the speed at which information reaches us means that we do not reach it. to contrast it, being the spearhead of fake news and not a problem of the AI itself.
Why is AI a potential danger to people?
Forget about Skynet or the Matrix, these are nothing more than human fantasies where racist concepts are transferred from the human species to another. After all, they are nothing more than alternative versions of the alien invasion and all of this is part of fantasy every day, so it is important to put your feet on the ground. We must start from the fact that the danger of AI is in turn its greatest virtue, since it is still a set of programs that learn about themselves as they receive new reference data with which to learn.
And that is precisely where the problems come from, specifically in four points that are important, but that depend in the first instance on those who introduce the data from which the AI is going to learn. That is, we are human beings who are going to manipulate the potential final conclusion. If one changes the premises, then the conclusion will. If I manipulate them to get a specific result, then the output will be invalid.
In other words, being an information verifier will be a job with a great future and that will require a specific and very strict professional profile if we do not want the dangers of AI to emerge.
Discrimination
One of the main problems with AI is that it cannot differentiate if the data is discriminatory towards a gender, ethnicity and sexual orientation. These inequalities, regardless of the sense in which we speak, can carry out totally disastrous decisions that can lead to discrimination of certain groups.
Privacy
Since large amounts of data are needed to make an AI work, a large part of the data that compromises people’s privacy ends up being used for it. Especially by large companies with large volumes of customers, internet search engines, social networks and even the government of your nation. The problem? That this implies additional protection measures to protect said information, something that many times, unfortunately, does not happen.
Another problem is that the companies themselves send texts to better document new products or to bring out design variations. This has recently happened to Samsung, which has found that its semiconductor department has used ChatGPT to get better text to document its products and has given it the option of using sensitive information.
Criminality
AI can provide solutions to very serious acts against people, such as giving an ideal route for a person on the run or how to hide a body after a murder. The big problem is that even though it understands natural language, there is to some extent the ability to confuse the AI by manipulating the input language and have it show a solution for a despicable act.
market manipulation
Many people ask the AI what product to buy, since it cannot test it itself, so look at all the websites where they talk about the product. And not only places where they do reviews professionally speaking, but also the enormous number that are in stores. Although the purchasing platforms can limit the comments to those who have the product, the company itself can buy a few units back as long as it can write false reviews in favor of or against the competition.
