Empathy

A machine can now produce the words. That is the change, and the reason this page needed rewriting.

Empathy is the capacity to understand another person's inner experience and respond to it, while holding the distinction between their experience and your own. Sympathy is feeling for somebody without necessarily understanding them. Compassion adds the motivation to act. Empathy is neither. Research separates the cognitive part, reading the other person's state, from the affective part, feeling some echo of it. Both do work, and someone with only one of them is recognisable.

What the evidence supports#

A study of nearly 7,000 managers across 38 countries found that those rated more empathetic by their subordinates also received higher performance ratings from their own superiors. Two different people, looking from two directions, agreeing.

In healthcare the evidence is unusually clean, and also more limited than it is usually made to sound. A systematic review pooled 28 randomised trials covering 6,017 patients. Seven of those trials tested empathic communication specifically, and across those seven the effect on pain, anxiety and satisfaction was small and real. The other twenty-one tested positive framing, which is a related but different intervention. That distinction disappears almost every time this review is cited, including in places that should know better.

What the evidence does not support#

Empathy is not stable, and the direction it has moved is not the flattering one. A cross-temporal meta-analysis of 72 samples of American college students found empathic concern down 48 per cent and perspective taking down 34 per cent between 1979 and 2009, most of the fall after 2000. Whatever is happening, it began before the tools.

Empathy is also not a substitute for judgement. Vivid feeling for one identifiable person reliably pulls resources away from larger but less visible needs. That is a measured effect and it does not make empathy a bad thing. It makes it a thing that needs calibrating against something else.

What AI changes#

Systems generate empathic language well, and in some comparisons responses written by a model have been rated more empathetic than responses written by a person, particularly when the person was overloaded.

The finding that matters sits next to it. Users rate machine-generated empathy as effective, and when given the choice they prefer to receive it from a human. The words can be right while the thing they are evidence of is absent, and people can tell the difference even when they cannot mark down the text.

The developmental risk is the one this research keeps returning to. Empathy is built by repeatedly attending to other people. If a system handles the ordinary interactions and a person is brought in only for escalations, the practice that builds the capacity is the part that was automated. That is cognitive offloading applied to a social skill, and it falls hardest on whoever is new, which is the same mechanism as the missing rungs.

What absence looks like#

Not coldness. Usually efficiency. Decisions that are defensible on their own terms and land badly for reasons nobody wrote down, followed by a surprise at the reaction.

In exit interviews it shows up as feeling unheard, which is reported more often than pay.

Practising empathy#

Two practices from SuperSkills (Kogan Page, 2026) carry this skill.

Run the sequence. Listen, label, ladder. Listen without preparing your reply. Label the feeling aloud so that something unspoken acquires language. Ladder by asking what would make this better, which is the step that turns understanding into a decision and the step most often skipped, because the first two improve the room and the third produces work.

Keep the three small rules. Small rules that scale: a three-second pause before replying, one felt metric recorded beside every performance goal, and a repair conversation within forty-eight hours of a rupture. Small enough to survive a difficult quarter, which is when practices of this kind are otherwise descheduled.

Run the third loop. The trust flywheel in the Five Loops closes with one decision a week recorded with a clear owner and a date visible to everyone affected.

Keep the everyday repetitions. The capacity develops through repeated practice of attending to others, so a person whose messages are always drafted for them is not practising. The organisational form of the same problem is a support model where machines handle first contact and humans meet only escalations, which removes the ordinary repetitions and leaves the hardest conversations to people who never built the skill.

What none of this establishes#

The sequence has not been tested as a sequence. Affect labelling has experimental support in psychology and that support does not transfer to the three steps taken together. The three-second figure is an observation from practice rather than a measured threshold.

Assessing empathy#

Organised by the ten subskills beneath this SuperSkill:

• Active listening, non-verbal interpretation. Whether they can report what the other person actually said, including the part that was inconvenient. Observable immediately and difficult to fake in a debrief.
• Emotional intelligence, interpersonal sensitivity. Whether labels get offered, and whether they are ever wrong. A person who always labels correctly is guessing safely.
• Conflict resolution and mediation, trust-building dialogue. Does a repair conversation happen after a rupture, and within how long? The forty-eight hour rule turns this into a countable thing.
• Storytelling for influence, clarity in digital communication. Whether the message changes when the audience changes, and whether the person can say what they altered and why.
• Cross-cultural communication, inclusive language and framing. Whether anybody outside their own context reviewed the work before it went out. The local mirrors roster is the practice that produces the evidence.

What this assessment cannot do#

Empathy is the skill where observation is most likely to reward performance. A person can run all three steps convincingly and change nothing, so the ladder step and the recorded decision matter more here than the conversation itself.

Key research and primary sources

• Sadri, G., Weber, T. J. and Gentry, W. A. (2011). Empathic emotion and leadership performance: An empirical analysis across 38 countries. The Leadership Quarterly, 22(5).
• Howick, J., Moscrop, A., Mebius, A. et al. (2018). Effects of empathic and positive communication in healthcare consultations. Journal of the Royal Society of Medicine, 111(7).
• Konrath, S. H., O'Brien, E. H. and Hsing, C. (2011). Changes in Dispositional Empathy in American College Students Over Time. Personality and Social Psychology Review, 15(2).

Every figure on this page has been checked against the source that reports it. Where a number could not be confirmed, it was removed rather than left standing, and what was removed is recorded in the build register.

About this research#

Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), keynote speaker on AI and human capability, and founder of The SuperSkills Intelligence Company.

How this research works · Reviewed quarterly · Found an error? Tell me and it is corrected on the page.

Listen. Heard on Education Futures, SuperSkills: The 7 human skills AI can't replace (with Rahim Hirji).

• Curiosity
• Change
  readiness
• Principled
  innovation
• Global
  adaptability
• Empathic
  communication
• Big
  picture thinking
• Augmented
  mindset