People Now Rate AI Empathy Higher Than Human Empathy.

In partnership with

Own Search With Podcasts

Your competitors are fighting over the same keywords. The smartest brands are building the authority that search engines, AI platforms, and customers trust everywhere.

Every relevant podcast appearance can produce branded mentions, backlinks, transcripts, citations, clips, expert content, and third-party proof that keeps compounding across search and AI discovery.

PodPitch searches millions of podcasts, finds the shows that matter to your market, develops the angle, sends personalized pitches, and follows up automatically until your experts are booked.

Growth teams are already using podcast appearances to build distributed authority that cannot be manufactured by publishing another generic SEO article.

Only 20 SEO, AEO, and GEO demo spots are available this month. Once they’re claimed, the offer disappears.

Start building searchable authority now, before your competitors own the conversations shaping your market.

Appear on 3,853,234 Podcasts

People Now Rate AI Empathy Higher Than Human Empathy.

Three studies published in recent years tested the same thing in different ways.

Researchers showed people two responses to an emotional situation. One written by a human. One written by AI. They asked which response felt more empathic.

The AI won. Consistently. Across different studies, different populations, different emotional scenarios.

That finding has been circulating for a while, usually framed as a curiosity or a warning about human emotional inadequacy. We are too distracted, too impatient, too poor at expressing care. AI does it better.

A paper published this year in Perspectives on Psychological Science reframed the finding in a way that changes what it actually means.

The author, Mark Thornton, a cognitive neuroscientist, introduced an analogy.

Placebos also work. People prefer them in some situations. They reduce pain. They produce measurable physiological benefits. The fact that they work is not in question.

The ethical consensus on placebos, despite their effectiveness, is that doctors should not be permitted to lie to patients to induce them. Not because the relief is not real. Because the mechanism that produces the relief requires deception. And deception, even in service of genuine benefit, has costs that the benefit does not cancel.

Thornton's argument is that AI empathy is the same thing. And the costs it carries are the ones worth understanding.

What AI Empathy Actually Is

When a large language model produces an empathic response, it is not feeling anything.

That point seems obvious. But its implications are not fully absorbed even by people who know it to be true.

Empathy in human relationships is not just a behavior. It is a behavior that signals an underlying state. When someone responds to your pain with genuine care, you are not just receiving a well-crafted sentence. You are receiving evidence that another mind has registered your experience and been affected by it. That evidence is what produces the sense of being understood. The words are the surface of something deeper that the words point to.

AI produces the words. It does not produce the underlying state the words are supposed to signal.

Thornton uses the wisdom-of-the-crowd analogy to explain why AI responses often outperform individual human ones. AI has been trained on the empathic responses of millions of humans. What it produces is, in effect, the statistical average of how humanity expresses care. The average of many humans, research consistently shows, outperforms most individual humans on judgment tasks.

So when AI beats human empathy in a head-to-head comparison, you are not watching a machine surpass human feeling. You are watching the averaged output of millions of human expressions of feeling outperform any individual human trying to express the same thing in the moment.

The feeling is humanity's. The delivery is the machine's. And the person receiving it does not always know which parts came from where.

The Moral Responsibility Shift

A paper published this year reviewing how AI reshapes human morality documented something precise about what happens when AI mediates emotional interactions.

People attribute less moral responsibility to situations when AI is involved in the decision or response. Not because they consciously decide to. Because the presence of AI changes the felt structure of the situation.

When a human responds to you, you hold them morally accountable for the response. When an AI responds to you, the accountability is diffused. The AI is not a moral agent. The company that built it is distant and abstract. The person who chose to use it is one step removed from the response you received.

That diffusion of accountability is not neutral. It changes how emotional exchanges are experienced and how their effects are processed.

A doctor who responds to a difficult diagnosis with cold efficiency can be held to account for the coldness. A patient can express that it hurt. The doctor can adjust. That feedback loop is part of how human relationships, including clinical ones, calibrate toward something better over time.

When an AI responds to the same diagnosis with carefully averaged warmth, there is no one to hold accountable if the warmth felt hollow. There is no feedback loop that adjusts human behavior. There is only the patient, alone with a response that was optimized to feel adequate and may have felt precisely that, adequate, and nothing more.

The CHI 2026 Finding on Empathy Mediation

Researchers presenting at CHI 2026 in Barcelona documented something that extends this further into everyday interaction.

AI-mediation, they found, shapes which affective cues are perceived in human interaction. When AI smooths emotional tone in text, when it filters the friction out of communication, when it translates anger into measured language or distress into calm requests, it is altering the emotional information that travels between people.

That alteration is not always bad. Smoothing the tone of an angry message before sending it can prevent escalation. Translating distress into clarity can make it easier to be heard.

But emotional information is not just noise to be managed. It is signal.

The frustration in someone's voice tells you something about how they are actually doing. The awkwardness in a difficult conversation tells you something about the stakes involved. The imperfect expression of care from a real person, halting and inadequate and trying anyway, tells you something about the relationship that a perfectly calibrated AI response cannot.

When AI systematically removes the friction from emotional communication, it is also systematically removing the information that friction carries. And the result is not cleaner communication. It is communication that is easier to send and harder to actually feel.

What Empathy Is Actually For

Here is the question the research is pointing toward but rarely states directly.

Empathy evolved as a relational and social capacity. Its function is not to make the recipient feel better, though that is a consequence. Its function is to create bonds between people. To signal that you are not alone. To build the trust that allows people to rely on each other across time.

That function requires reciprocity. It requires that the thing expressing care is itself capable of being affected by you in return. That the relationship involves mutual vulnerability and mutual investment.

AI can produce the signal of empathy. It cannot participate in the relationship that empathy is supposed to be building.

When people increasingly receive their emotional needs met by systems that cannot be in relationship with them, the need gets addressed in the short term and the capacity for human relationship atrophies in the long term. Not dramatically. Not in a way that is immediately visible. In the same slow, accumulating way that all the other attritional effects of AI on human capacity work.

The Placebo Consensus and What It Means

Thornton ends his paper with a question that the field has not answered.

If AI empathy is a placebo, the ethical framework that governs placebos suggests that its use should be constrained by transparency. Patients who receive placebos should know they are receiving them. Not because the effect disappears when you know. But because the deception required to conceal it has costs to the relationship and to trust that the benefit does not cancel.

The same question applies to AI empathy. When someone receives a response that is designed to feel empathic from a system that is incapable of empathy, do they know? Are they choosing that? Or is the mechanism designed to be invisible precisely because visibility would reduce the effect?

That question is not resolved. But it is worth asking before you decide how much of your emotional life to route through systems that were never designed to be in relationship with you.

If something in this landed closer to your actual experience than you expected, that recognition is worth something. Send it back if you want to think it through with someone who is actually here.

If this newsletter makes you think differently each week, one click on the ad above is the easiest way to keep it going. No subscription, no paywall. Just that one small thing.

Thanks for reading. See you Tuesday.

The AI bill isn't what sinks you.

AI costs don't usually explode overnight. They grow quietly through duplicate requests, expensive routing and poor visibility.

Mesh API helps engineering teams spot the waste before finance does.

A global e-commerce company was able to reduce spends by 78%.

Estimate Savings