■ MODERATE RISK ■ Healthcare
No. Continuous monitoring and risk prediction are being automated aggressively, but birth is a physical, unpredictable, deeply relational event, and midwifery is the profession of being present through it.
“AI monitors labor, but someone still needs to say 'push' convincingly.”
Our AI replacement risk score — how we score jobs
Midwifery spans far more than the delivery room. There is antenatal care — booking appointments, blood pressure and urine checks, palpation, growth measurement, screening discussions, mental health assessment, and the slow work of learning what a specific woman is frightened of. There is labour: monitoring fetal heart rate and contractions, managing pain, positioning, deciding when a slow progress is fine and when it is the first sign of trouble, escalating to obstetrics, and often catching the baby. Then postnatal care, feeding support, safeguarding assessments, and community visits. Documentation follows everything, endlessly.
Machine input is already routine on the monitoring side. Cardiotocograph interpretation is a classic pattern-recognition problem where algorithms reduce the variability between human readers; risk-scoring models flag pre-eclampsia, gestational diabetes and preterm risk from routine data; wearable monitors extend surveillance into the home. Scheduling, triage lines and antenatal education are increasingly handled by apps and chatbots. Ambient documentation tools are starting to remove the note-writing burden that dominates the shift. These are meaningful changes, and they will change staffing ratios in monitoring-heavy settings.
The core of the job resists for reasons that are not sentimental. Labour requires hands: palpation, perineal support, manoeuvres for shoulder dystocia, resuscitation of a flat newborn, controlling a postpartum haemorrhage in the ninety seconds before help arrives. It requires continuous judgment in a rapidly changing situation with two patients whose interests can diverge. It requires trust deep enough that a woman will disclose domestic abuse or drug use. And it carries professional accountability for decisions made alone at 4am in someone's living room. Our risk score of 28 reflects heavy technological augmentation around a role that stays physically and ethically human.
Automatability: our editorial assessment of current and near-term AI capability
Algorithmic CTG interpretation and risk stratification are entering routine practice now and will be standard across developed systems within this decade. Documentation relief follows close behind. Neither reduces the need for hands at a birth, so through the 2030s the likely effect is redistributed workload — less time watching screens, more time in direct care — with the profession still chronically short-staffed rather than shrinking by 2040.
Not in any near-term sense. Delivery is manual work in an unpredictable, fluid-covered environment with two patients and a constant possibility of emergency. Even the most advanced surgical robots operate on a fixed field under direct human control. The monitoring and paperwork around birth automate readily; the birth itself does not.
One of the more secure in healthcare. Demand is demographic, regulation requires registered practitioners, and the work is physical, relational and emergency-prone — the three qualities automation handles worst. The realistic pressures on midwifery are funding, staffing and burnout, which existed long before anyone mentioned machine learning.
Mostly in surveillance and triage: algorithmic interpretation of fetal monitoring, risk prediction models built on booking data, remote blood pressure monitoring for hypertensive pregnancies, and chatbots handling routine antenatal questions. Increasingly, ambient tools also draft clinical notes, which is the change most midwives would vote for first.
Clinical depth first — advanced practice qualifications, ultrasound, prescribing, emergency skills. Then digital literacy sharp enough to challenge a monitoring algorithm's output when it conflicts with what your hands and eyes are telling you. Documented, confident overriding of a flawed tool is becoming a defining professional competence.