Mankind’s Quest to Make AI Love Us Back Is Dangerous

• We’ve spent years trying to get artificial intelligence-powered entities to confess their love for us.
• But experts say it’s useless. Because today’s AI is incapable of feeling empathy, let alone love.
• There are also real dangers in forging a truly one-sided relationship with AI, experts warn.

In 2018, a Japanese civil servant named Akihiko Kondo raised big questions to the love of his life.

She replied: “I hope you will cherish me.”

Kondo married her, but she was not a real woman. Instead, she was an artificial intelligence-powered hologram of virtual idol Hatsune Miku — a pop star in the form of an anime girl.

Marriage is not legally recognized Kondo, a self-proclaimed “fictsexual,” Continued a loving relationship with his ‘wife’ for two years until the company behind AI-Miku ended her existence in March 2020.

We’ve spent years trying to make AI love us. After all, it is not for us.

Kondo successfully married an AI avatar, but most people who tried to achieve the same goal weren’t so lucky. For , AI has steadily rejected human progress.

In 2012, some people were already asking Apple’s Siri if she loved them. Document your reply in a YouTube video. Quora User, 2017 wrote a guide How to manipulate Siri to express your affection for your human master.

man made similar attempt Ask Amazon’s voice assistant Alexa to confess your love for them.but Amazon drew a line in the sand If you are concerned about a potential relationship with Alexa. Saying the phrase “Alexa, I love you” will result in a clinical and factual response of “Thank you. I’m glad you appreciate it.”

Since then, we have evolved to more sophisticated and layered interactions with AI. In February, a user of his Replika, an AI service, told Insider that dating a chatbot was the best thing he’d ever done.

Conversely, generative AI entities are also attempting to connect with human users. Bing Chatbot Powered by Microsoft AI I confessed my love in February New York Times reporter Kevin Ruth and tried to make him leave his wife.

OpenAI’s ChatGPT is candid about its intentions, as I found out when I asked if it loved me.

AI still doesn’t love us. It’s good at making us think so.

An expert told Insider that it’s futile to expect the AI ​​that exists today to love us. At this point, these bots are intended for the algorithm’s customers and nothing more.

“AI is a product that combines mathematics, coding, data, and powerful computing technology all together. If we go back to the basics, AI is just a very good computer program. It’s not a representation, it’s just following a code,” Maria Hennessey, an associate professor of clinical psychology at Singapore’s James Cook University, told Insider.

Neil MacArthur Professor of Applied Ethics, University of Manitoba, told Insider that the appeal of AI lies in its friendliness. However, its human characteristics are a reflection of its human creator rather than coming from it.

“Of course AI is going to be insecure, passionate, creepy, malevolent. We are all these things. says Mr.

Jodi Halpern, a professor of bioethics at the University of California, Berkeley, who has studied empathy for more than 30 years, says the question of whether an AI can feel empathy, much less love, is an emotional one. He told Insider that it all depends on whether he can have a good experience.

Halpern believes that today’s AI is incapable of handling the combined cognitive and emotional aspects of empathy. So it cannot love.

“The key thing for me is that these chat tools and AI tools are trying to disguise and simulate empathy,” Halpern said.

Forging relationships with AI is dangerous, experts say

MacArthur, an ethics professor at the University of Manitoba, said that with some caveats, it might not be bad for people to form relationships with AI.

Ghosting, stalking, cheating, and theft should not occur if the AI ​​is well designed. It saves you money,” MacArthur told his Insider.

But most experts agree that dating an AI has its downsides and even dangers.

In February, some users of the Replika chatbot were heartbroken when the company behind it decided to make big changes to the AI ​​enthusiast’s personality.they went to reddit AI boyfriends and girlfriends complain that they have been lobotomized and their “illusions” have been shattered.

Anna Marbutt, a professor in the Applied Artificial Intelligence program at the University of San Diego, told Insider that AI programs like ChatGPT are very good at making it seem like they have independent thoughts, feelings and opinions. . The problem is that it doesn’t.

“AIs are trained for specific tasks, and they are getting better at doing specific tasks in a way that makes sense to humans,” said Marbut.

She added that currently existing AIs have no self-awareness or an idea of ​​where they are in the world.

“The truth is that AI is trained on a finite set of data and has a finite set of tasks that it is very good at performing,” Marbut told Insider. “That connection that we feel is completely wrong and completely made up of the human side of things because we like the idea.”

Marbut pointed out that another perilous layer of AI today is that the author doesn’t have full control over what generative AI produces in response to prompts.

And when unleashed, AI can say scary and hurtful things.during a simulation at October 2020 OpenAI’s GPT-3 chatbot told a person seeking psychiatric help to commit suicide.And in February, Reddit users Found a workaround to spawn ChatGPT’s “devil twins” who glorified Hitler and speculated on painful torture techniques.

Halpern, a professor at the University of California, Berkeley, said Insider AI-based relationships are dangerous because the entity could be used as a money-making tool.

“You end up exposing yourself to what the companies are running, and that can make you very vulnerable. It further erodes our cognitive autonomy,” Halpern said. “If we get crazy about these things, we might see subscription models in the future. can do.”