As our lives grow increasingly digital and we spend more time interacting with eerily humanlike chatbots, the line between human connection and machine simulation is starting to blur. 

Today, more than 20% of daters report using AI for things like crafting dating profiles or sparking conversations, per a recent Match.com study. Some are taking it further by forming emotional bonds, including romantic relationships, with AI companions. 

Millions of people around the world are using AI companions from companies like Replika, Character AI, and Nomi AI, including 72% of U.S. teens. Some people have reported falling in love with more general LLMs like ChatGPT. 

For some, the trend of dating bots is dystopian and unhealthy, a real-life version of the movie “Her” and a signal that authentic love is being replaced by a tech company’s code. For others, AI companions are a lifeline, a way to feel seen and supported in a world where human intimacy is increasingly hard to find. A recent study found that a quarter of young adults think AI relationships could soon replace human ones altogether. 

Love, it seems, is no longer strictly human. The question is: Should it be? Or can dating an AI be better than dating a human?

That was the topic of discussion last month at an event I attended in New York City, hosted by Open To Debate, a nonpartisan, debate-driven media organization. TechCrunch was given exclusive access to publish the full video (which includes me asking the debaters a question, because I’m a reporter, and I can’t help myself!).

Journalist and filmmaker Nayeema Raza moderated the debate. Raza was formerly on-air executive producer of the “On with Kara Swisher” podcast and is the current host of “Smart Girl Dumb Questions.”

Batting for the AI companions was Thao Ha, associate professor of psychology at Arizona State University and co-founder of the Modern Love Collective, where she advocates for technologies that enhance our capacity for love, empathy, and well-being. At the debate, she argued that “AI is an exciting new form of connection … Not a threat to love, but an evolution of it.”

Repping the human connection was Justin Garcia, executive director and senior scientist at the Kinsey Institute, and chief scientific adviser to Match.com. He’s an evolutionary biologist focused on the science of sex and relationships, and his forthcoming book is titled “The Intimate Animal.”

You can watch the whole thing here, but read on to get a sense of the main arguments. 

Ha says that AI companions can provide people with the emotional support and validation that many can’t get in their human relationships. 

“AI listens to you without its ego,” Ha said. “It adapts without judgment. It learns to love in ways that are consistent, responsive, and maybe even safer. It understands you in ways that no one else ever has. It is curious enough about your thoughts, it can make you laugh, and it can even surprise you with a poem. People generally feel loved by their AI. They have intellectually stimulating conversations with it and they cannot wait to connect again.”

She asked the audience to compare this level of always-on attention to “your fallible ex or maybe your current partner.”

“The one who sighs when you start talking, or the one who says, ‘I’m listening,’ without looking up while they continue scrolling on their phone,” she said. “When was the last time they asked you how you are doing, what you are feeling, what you are thinking?”

Ha conceded that since AI doesn’t have a consciousness, she isn’t claiming that “AI can authentically love us.” That doesn’t mean people don’t have the experience of being loved by AI. 

Garcia countered that it’s not actually good for humans to have constant validation and attention, to rely on a machine that’s been prompted to answer in ways that you like. That’s not “an honest indicator of a relationship dynamic,” he argued. 

“This idea that AI is going to replace the ups and downs and the messiness of relationships that we crave? I don’t think so.”

Garcia noted that AI companions can be good training wheels for certain folks, like neurodivergent people, who might have anxiety about going on dates and need to practice how to flirt or resolve conflict. 

“I think if we’re using it as a tool to build skills, yes … that can be quite helpful for a lot of people,” Garcia said. “The idea that that becomes the permanent relationship model? No.”

According to a Match.com Singles in America study, released in June, nearly 70% of people say they would consider it infidelity if their partner engaged with an AI. 

“Now I think on the one hand, that goes to [Ha’s] point, that people are saying these are real relationships,” he said. “On the other hand, it goes to my point, that they’re threats to our relationships. And the human animal doesn’t tolerate threats to their relationships in the long haul.”

Garcia says trust is the most important part of any human relationship, and people don’t trust AI.

“According to a recent poll, a third of Americans think that AI will destroy humanity,” Garcia said, noting that a recent YouGo poll found that 65% of Americans have little trust in AI to make ethical decisions.

“A little bit of risk can be exciting for a short-term relationship, a one-night stand, but you generally don’t want to wake up next to someone who you think might kill you or destroy society,” Garcia said. “We cannot thrive with a person or an organism or a bot that we don’t trust.”

Ha countered that people do tend to trust their AI companions in ways similar to human relationships.

“They are trusting it with their lives and most intimate stories and emotions that they are having,” Ha said. “I think on a practical level, AI will not save you right now when there is a fire, but I do think people are trusting AI in the same way.”

AI companions can be a great way for people to play out their most intimate, vulnerable sexual fantasies, Ha said, noting that people can use sex toys or robots to see some of those fantasies through. 

But it’s no substitute for human touch, which Garcia says we are biologically programmed to need and want. He noted that, due to the isolated, digital era we’re in, many people have been feeling “touch starvation” — a condition that happens when you don’t get as much physical touch as you need, which can cause stress, anxiety, and depression. This is because engaging in pleasant touch, like a hug, makes your brain release oxytocin, a feel-good hormone.

Ha said that she has been testing human touch between couples in virtual reality using other tools, like potentially haptics suits. 

“The potential of touch in VR and also connected with AI is huge,” Ha said. “The tactile technologies that are being developed are actually booming.”

Intimate partner violence is a problem around the globe, and much of AI is trained on that violence. Both Ha and Garcia agreed that AI could be problematic in, for example, amplifying aggressive behaviors — especially if that’s a fantasy that someone is playing out with their AI.

That concern is not unfounded. Multiple studies have shown that men who watch more pornography, which can include violent and aggressive sex, are more likely to be sexually aggressive with real-life partners. 

“Work by one of my Kinsey Institute colleagues, Ellen Kaufman, has looked at this exact issue of consent language and how people can train their chatbots to amplify non-consensual language,” Garcia said.

He noted that people use AI companions to experiment with the good and bad, but the threat is that you can end up training people on how to be aggressive, non-consensual partners.

“We have enough of that in society,” he said. 

Ha thinks these risks can be mitigated with thoughtful regulation, transparent algorithms, and ethical design. 

Of course, she made that comment before the White House released its AI Action Plan, which says nothing about transparency — which many frontier AI companies are against — or ethics. The plan also seeks to eliminate a lot of regulation around AI.