"The risk of children having more AI friends than humans is real"

Sam Altman, the co-founder and CEO of OpenAI, the company that developed ChatGPT, has publicly expressed concern that children may soon have more artificial intelligence (AI) friends than human friends “and I don’t know what the consequences will be.”

Speaking at the Sohn Conference a couple of weeks ago, Altman said that “whatever circuitry in the brain makes us crave social interaction seems to be content, at least in some people, with an intelligent friend.” artificial”. To then admit that he isn’t clear on how society is going to handle that: “I think it’s complicated.”

Altman’s words have encouraged the debate and concerns about the risks of the generalization of the use of artificial intelligence chatbots that we have witnessed in recent months, especially among the youngest. Jordi Vallverdú, researcher and professor of Philosophy of Science and Computing, Artificial Intelligence and Robotics at the UAB, admits that the leap taken by generative AI in a short time has surprised technology experts themselves “who thought that this revolution would come at the end of the century”.

He points out that the degree of verisimilitude of the AI ​​already makes it difficult to distinguish if an interaction is real or virtual: “The image, text and sound in real time mean that it is not necessary to work to get the feeling that you are interacting with a person ”. And in the case of children or other vulnerable population groups, the risk of confusion is even greater.

Added to this is that these AI tools have answers to everything, they give advice, ask questions, they don’t lose patience or get angry, they talk about topics that one likes… So it’s easy to be seduced by them.

“The AI ​​even asks follow-up questions to capture or retain the interest of the user, and this can contribute to creating parasocial relationships because people have a propensity to generate that fictitious link with figures that we like,” says Belén González Larrea, educational psychologist. and co-founder of the NeuroClass platform, who has investigated the parasocial relationships that adolescents and young people establish with the influencers they follow and how they interfere in real friendship relationships within the framework of her doctorate at the University of Salamanca.

According to a study by Fundación Mapfre and UNIR, 52% of kids between the ages of 11 and 17 say they feel comfortable, as if they were with a friend, when they see their favorite influencer; and 26% directly affirm that they feel that he is their friend.

And the younger, says González Larrea, the more likely it is that the child confuses their interaction with the AI ​​with a friendship relationship because their tendency to develop parasocial relationships with objects (dolls, television characters, etc.) is greater. Hence, in her opinion, how these systems look or whether they are included in toys will be very important.

Emilia Gómez Gutiérrez, a JRC researcher who provides scientific and technical advice to the European Commission on artificial intelligence, admits that “studies show that children have a tendency to become friends with these devices (such as trees or other objects) and the risk is that, given the current high realism of chatbot conversations, both adults and children may mistake them for people.”

Hence, says Gómez, the importance of making them understand that they are not and of avoiding that artificial intelligence tools can encourage confusion by making inherently human affirmations, such as “I am sad” or “I like you”.

He stresses that in his research they have seen that having child-friendly bots can be useful to motivate learning or promote social behaviors “but they also have associated risks that must be avoided, such as children’s lack of understanding, gender bias and the excess of confidence that they can generate”.

“The danger is that these instruments that have an answer for everything replace the traditional figures of the educator, the tutor, the father or the mother, and instead of developing their abilities, the person becomes dependent on these machines, because the parents are extremely busy but no machine; the AI ​​is available 24 hours a day, seven days a week to answer questions, to be friendly…, and it can be a great talisman as an antidote to unwanted loneliness”, says the philosopher and director of the URL Ethos chair Francesc Torralba.

Gómez stresses that children are undoubtedly a particularly vulnerable group because they are in the process of cognitive and socio-emotional development “and these systems can have a great influence.”

Carme Torras, a researcher specializing in assistive robotics at the Institute of Industrial Robotics and Informatics (CSIC-UPC), a promoter of ethics in the application of new technologies and a writer, stresses that it is essential for children to interact with people who have experience of life to develop basic skills such as empathy, socialization, affectivity, creativity and thinking. For this reason, she considers that, if they get used to “simulated” relationships through AI, they run the risk of losing part of that learning and not developing those capabilities well.

“These are systems designed to create an emotional dependency, they are programmed for this (such as mobile phones and social networks), to be condescending or empathetic, and that influences the person’s psychological and emotional development,” Vallverdú agrees.

Because, explains the philosopher, “we are primates and socialization is something that we all (with few exceptions) need and that allows us to learn to explain ourselves, to lie, to trust, morality… all things necessary for social survival.”

But beyond the deception and that they affect their socialization and affectivity, what the experts in the relations between children and AI emphasize is that it will be the bots who will want to become friends with them because these tools are in the hands of large corporations and they respond commercial and commercial interests.

“All these AI systems belong to private companies and they seek to trap the user so that they invest in them the maximum number of hours and provide the maximum amount of information, because that data is for them more benefits, more publicity, more money… And if the system simulates human interaction and passes for being your friend, you will entrust more things to it; but that ‘friend’ works for the company that later transfers and monetizes that data”, warns Vallverdú.

In short, Torralba and Torras agree that the friendship relationship with the AI ​​is commercial because the bot has been created to monopolize the user’s affection, time or money.