Article image

AI outperforms humans in providing emotional support

Feeling truly heard and understood is a rare thing these days. Life moves quick, and many of us don’t get the emotional support, empathy and validation we need from those around us. So, could artificial intelligence (AI) do it? A new study conducted at the USC Marshall School of Business says yes – with some intriguing caveats.

AI, a surprisingly good listener

Researchers found that people who received messages generated by AI actually felt more heard than those receiving messages written by a human!

The AI also excelled at detecting the person’s emotions. This might seem counterintuitive, but there are a few key reasons that might explain this surprising result:

AI systems are specifically designed to analyze language and identify emotional cues. They don’t get distracted by their own thoughts, feelings, or the myriad of other things going on in their day-to-day lives the way humans can. This allows them to dedicate more processing power to understanding the emotional message being conveyed.

AI systems (ideally) operate from a neutral starting point, without the human tendency for preconceived notions or unconscious biases. Approaching each message from a fresh perspective helps in recognizing the emotions as they are expressed.

AI models are trained on massive datasets of text and human emotional exchanges. This gives them an extensive library of patterns and language cues to draw from, allowing them to detect subtle nuances in emotional communication that a layperson might miss.

AI offers the right kind of emotional support

It might come down to a surprising advantage AI has over humans. Computers are programmed to analyze and categorize, free of all the distractions that plague our human brains.

“In the context of an increasing loneliness epidemic, a large part of our motivation was to see whether AI can actually help people feel heard,” explained study’s lead author, Yidan Yin.

The study indicates that the AI-generated messages were better at focusing on acknowledging and validating the recipient’s emotional state. This created a sense of being heard and seen, boosting hope and lessening distress.

Moreover, AI may be better at recognizing when simple emotional support is what’s required, rather than jumping in with solutions. This ability to discern the right type of response likely contributed to its positive impact.

AI sticks to validation

AI’s lack of personal experience and emotional biases allows it to stick to pure emotional reflection without interjecting its own opinions or judgment. This non-judgmental approach creates a safe space for the person to feel fully understood.

Humans often try to “fix” someone’s problems with practical advice when a person is primarily seeking emotional support. AI doesn’t have this impulse, which can help the person feel truly heard and free to process their emotions without the pressure of finding a solution.

The problem with emotional support of AI

But, and this is a big “but,” here’s where things get strange. When participants found out their empathetic messages came from AI, they actually felt less heard. This phenomenon is similar to the “uncanny valley” effect in robotics, where humans get uncomfortable when something is almost human, but not quite.

“What we found was that both the actual source of the message and the presumed source of the message played a role,” says co-author Cheryl Wakslak. “People felt more heard when they received an AI than a human message, but when they believed a message came from AI this made them feel less heard.”

Humans vs. AI: It’s not an “either-or”

This research doesn’t mean AI is going to replace our best friends as the best shoulder to cry on. Rather, it points to a fascinating collaboration potential.

Humans could learn a lot from how AI analyzes emotions and tailors its language for validation. “Humans may potentially learn from AI because a lot of times when our significant others are complaining about something, we want to provide that validation, but we don’t know how to effectively do so,” notes Yin.

Moreover, AI solutions could provide accessible and affordable emotional support, particularly for those who lack other outlets or social resources.

Feeling heard via AI’s emotional support

This research demonstrates that AI has the potential to understand our emotions and offer validating responses in ways that can be incredibly beneficial.

This opens up possibilities for accessible mental health support, tools for better communication, and perhaps even deeper self-understanding.

The “AI Uncanny Valley” effect, where people are unsettled by something that’s almost human but not quite, is a significant obstacle.

The knowledge that AI generated their supportive messages made participants feel less understood. This suggests a psychological barrier we’ll need to overcome.

Becoming more comfortable with AI

It makes you wonder, with the inevitable advance of AI, will our attitudes soften? Will the “AI Uncanny Valley” effect fade as we get used to computers that truly understand how we feel?

These are pivotal questions for the future of human-AI interaction. It’s possible that with exposure and positive experiences, our discomfort with AI in emotional contexts will lessen.

We might begin to view it less as a replacement for human connection and more as a unique tool with its own set of strengths.

However, it’s also possible that the unease caused by the “uncanny valley” effect will persist, posing a fundamental challenge to widespread adoption of AI in roles that require deep emotional understanding.

Study significance

As co-author Nan Jia puts it, “we identified that while AI demonstrates enhanced potential compared to non-trained human responders to provide emotional support, the devaluation of AI responses poses a key challenge for effectively deploying AI’s capabilities.”

The idea of an AI therapist might still be a ways off, but this study proves something important: AI has the potential to understand and respond to our emotions in ways we’re still figuring out. It’s a tool, and a powerful one. Whether we end up using it to have deeper connections with computers or with each other…that remains to be seen.

The study is published in Proceedings of the National Academy of Sciences of the United States of America.


Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and


News coming your way
The biggest news about our planet delivered to you each day