The Hardest Conversation in Fertility Medicine Has No Algorithm
AI has learned to respond with warmth when a patient asks if her cancer will kill her. The question is not asking for warmth. It is asking for something no algorithm can give.
The two-week wait after an embryo transfer is one of the most psychologically intense periods in reproductive medicine. Patients know the statistics. They have read the SART data. They have calculated their clinic’s success rates by age, by diagnosis, by transfer type. And then they wait. The knowing does nothing to protect them.
Reproductive endocrinology and infertility has a particular relationship with hope that no other subspecialty quite replicates. Patients arrive having already processed loss: miscarriages, failed cycles, diagnoses of diminished ovarian reserve, partner factor infertility, unexplained infertility that is its own particular cruelty. They arrive sophisticated about their own bodies. And they arrive needing something that the AI empathy debate has consistently failed to name correctly.
What they need is not empathy. What they need is compassion.
The distinction between those two words is the argument this post makes, and it matters most in REI because this subspecialty has a unique capacity to confuse one for the other, with real consequences for patients.
Empathy Is Not Enough
Empathy is the perception of another person’s emotional state and some resonance with it. It can be trained, and as recent research confirms, it can be convincingly simulated. Patients shown AI chatbot responses to medical questions rated them as more empathic than physician responses. That finding has been replicated across settings.
Compassion goes further. It is empathy plus the moral commitment to act on what you perceive, including when that action is unwelcome. The compassionate REI clinician does not just recognize her patient’s grief and hope across multiple failed cycles. She is moved by it to do something: to explain, to guide, to stay, and, when the evidence warrants it, to tell her patient honestly that continuing treatment is unlikely to change the outcome. That last act is what communicative empathy cannot perform. It requires a clinician who cares enough about the patient’s whole life to disappoint her.
Bioethicist John Lantos frames the distinction as communicative empathy versus moral empathy. Large language models have mastered the first. What they cannot do is the second: a genuine responsiveness to another person’s suffering that reshapes how you act. In REI, this distinction is not philosophical. It is where clinical harm begins.
— PAYWALL BREAK — Free readers see above. Paid subscribers continue below.
