Will AI replace Physical Therapists?

Overall automation risk is very low because the core value of physical therapy relies heavily on tactile evaluation, manual interventions, and real-time physical coaching. Administrative duty types such as chart reviews, progress notes, and routine patient education materials exhibit the highest vulnerability to automated workflow assistants. Clinicians should pilot ambient AI scribing software this quarter to cut documentation time and redirect focus entirely toward direct patient interaction.

Partially. Physical Therapists scores 35/100 — some core duties are highly automatable, but enough durable human work remains that the occupation is transforming rather than disappearing overnight. This is a task-exposure index, not a guarantee that hiring stops.

The score is an importance-weighted average of automation probabilities across the top 15 O*NET tasks for SOC 29-1123.00. 1 task score at or above 80% automatable; 8 fall into the safer band (under 30% or labeled physical). Roughly 27% of scored tasks are primarily digital.

None of the top 15 tasks currently clear the ≥80% automation threshold — exposure is more diffuse across mid-range probabilities.

Few tasks in this profile clear the “safe” threshold, which is why the aggregate score skews higher and why adjacent lower-risk careers deserve serious consideration.

Official BLS data places median pay for this occupation family at $102,760. with projected employment change of +11.9% over the latest 10-year outlook window. Typical entry education: Doctoral or professional degree. Wage and growth context ($102,760, +11.9%) should be read alongside the AI score — not as a substitute for it.

Physical therapists should master ambient clinical documentation and AI-driven diagnostic synthesis tools to significantly reduce electronic health record overhead. Deepening specialized manual therapy skills, complex neuromuscular rehabilitation, and interdisciplinary care coordination will further insulate practitioners against technological displacement. Additionally, learning to supervise digital home-exercise adherence platforms will allow therapists to expand patient capacity without compromising care quality.

We pull Core O*NET task statements for Physical Therapists, score each for Generative AI automation probability, weight by O*NET importance, and merge the result with BLS wages and employment projections on the SOC code. Full methodology, limitations, and prompt versioning are documented on the methodology page.

Physical Therapists possesses robust structural insulation (80/100, verdict: "High Physical/Social Insulation"). Large language models and automated software copilots operate exclusively within digital runtimes. Because this role demands significant physical proximity (98/100), direct interpersonal presence (95/100), and psychomotor coordination (49/100), it remains heavily defended against pure software substitution. Physical Proximity & On-Site Presence is the primary barrier (98/100), protecting human workers from algorithmic displacement.

Federal OEWS data reveals an earning spread of $58,610 from the 10th percentile ($72,260) to the 90th percentile ($130,870). The middle 50% of practitioners earn between $81,970 and $113,630. Compensation for Physical Therapists reflects a hybrid balance of cognitive judgment and physical/interpersonal execution. Top-tier compensation ($130,870) is driven by complex problem-solving and domain mastery that resists routine software automation.

Academic research from OpenAI and UPenn strongly aligns with our AI Career Stats assessment. Both algorithmic task evaluations (Gemini 3.8 Flash Task Model: 35/100, GPT-4 direct exposure: 9/100) and human expert panels (34/100) arrive at a shared consensus on the automation trajectory for Physical Therapists. Software tooling expansion increases exposure by +10 points (from 9/100 to 19/100), demonstrating that integrating AI into existing software suites significantly expands automated task throughput.