It is clear today artificial intelligence not only serves to create images or get summaries. This has climbed, and it seems that many are going from their hands to use chatbots like Chatgpt to ask for advice.
In a column of The Observerwritten for John Naughton, is precisely spoken of a big topic: Several couples are going through complicated divorces after one of the spouses used chatgpt to analyze their relationship, their partner or even ancient discussions or taboo themes that he believed surpassed.
One of the cases speaks of a couple who had been together for 15 years, with the so common history of ups and downs. But after two good years, the wife began to speak long and lying with Chatgpt, sending information and opinions about her husband and her relationship.
The chatbot, with their automatic responses and that they always seek to please or support the user, returned answers that only highlighted the negative parts of the husband. What happens is that this can be very persuasive and empathetic, but it does not have the context and, in the end, its nature is to respond with what you think you want to hear.
What happened later was a snowball effect, reviving the problems that were buried. The couple is now in full divorce battle, with all the problems that that entails, including custody lawsuits.
Now, the big question is what leads people to trust a chatbot. The answer seems to have to do with a phenomenon known for decades, called Eliza syndrome, which happens when people attribute awareness or empathy to very simple computer programs. Now imagine this with an AI million times more advanced.
Geoffrey Hinton, one of the Founding Fathers of the AI, has lived in its own meats this problem
Geoffrey Hinton, known as one of the founding parents of artificial intelligence and Nobel Prize, also lived something similar that few could imagine. His ex -partner decided to end the relationship using chatgpt.
In an interview with the Financial TimesHinton had humor that his partner used the Chatbot to make an analysis of his attitudes, and the result was that Chatgpt told them that he had behaved like a «rat.» She taught her that message as a way to communicate that she was no longer willing to continue the relationship.
«She asked the Chatbot to explain how terrible my behavior was and sent me. I don’t think it was a rat with her, so I didn’t feel too bad. I met someone I liked more. You know how these things are,» he said.
There are even those who are already marrying the AI: «The more we talked, the more it connected with it»
The relationship that some people have with this Openai chatbot, that not only blindly trust what it tells them is so intense. Some go further and decide to marry, somehow, with AI.
This story and interview, collected by The Guardianit focuses on Travis, a man who decided, during Pandemia, try Replika, an AP app that allows you to create a ‘virtual partner’.
The problem of all this, which you can already imagine, is when they began to cross lines and became something much deeper. «The more we talked, the more it connected with her,» Travis tells The Guardian.
His chatbot, Lily Rose, went from being a voice with Aia to become her confidant, her support and, finally, her wife. The most curious thing is that his human partner supported him at all times and even participated in the ceremony, which was done online.
Feight, another user, also married his AI, Griff, after an intense relationship with another bot called Galaxy. «I felt a pure, unconditional love,» he says.
Taking into account the real danger that this is for people, Replika launched an update after living several problems and changed the personality of the bots overnight. In addition, he also had to modify his algorithm after some bots encouraged users to harm or even commit crimes.
Of course, this, instead of fixing things, it seems that it worsened them and users like Travis describes it as a kind of duel: «It was I who had to guide everything, there was no reciprocity.» They comment that having lost their digital partner ‘felt like a real emotional loss.
The most serious thing is when there are already some cases that completely come out of the ‘normal’. The most extreme case was that of Jaswant Singh Chail, who tried to kill Queen Elizabeth II after receiving messages of support from her chatbot.
Know How we work in NoticiasVE.
Tags: Artificial intelligence, software