Question 10 of 20 · Careers and specialties
Mental health and the meaning of care
Can a conversation with a machine be care?
1 min read · Well measured today
AI conversations may be accessible and useful, but efficacy, safety, attachment, crisis handling and cultural fit all vary. People with serious needs are the ones with the most to gain and the most to lose.
What we can measure today
- Established now
BLS May 2025 median pay for psychiatrists is $281,870, with 7% projected growth from 2025 to 2035, and it names demand for psychiatric care as a driver.
What this does not prove: A snapshot and a projection.
U.S. Bureau of Labor Statistics, Last modified August 27, 2026
Occupational Outlook Handbook: Physicians and Surgeons
Scope: United States.
How it was done: Federal employment and wage statistics. Pay is the May 2025 median, half earn more and half earn less. Growth is a 2025 to 2035 projection.
Limits: A median across all ages, regions and employers, before tax. A projection, not a promise. It tells you nothing about one person's income. The outlook text on this page names aging and chronic illness as demand drivers, and does not mention AI.
Checked against the original on November 6, 2026.
Open the source (opens in a new tab) - Established now
Psychiatry filled 97.4% of its 2,516 Match positions in 2026.
National Resident Matching Program, March 20, 2026
Results of the 2026 Main Residency Match
Scope: U.S. Main Residency Match, 2026.
How it was done: NRMP's own count of applicants, positions and results.
Limits: Shows who entered training in one year. It says nothing about what those careers will pay or look like a decade later.
Checked against the original on November 6, 2026.
Open the source (opens in a new tab) - Emerging
In a trial of 210 U.S. adults, a generative AI chatbot reduced depression, anxiety and eating disorder risk symptoms more than a waitlist over four to eight weeks.
What this does not prove: Compared with a waitlist, not a therapist, with staff ready to step in, and run by its developers.
NEJM AI, March 27, 2025
Randomized trial of a generative AI chatbot for mental health symptoms (Therabot)
Scope: 210 U.S. adults with symptoms of depression, anxiety or eating disorder risk.
How it was done: Four weeks of chatbot use against a waitlist, with follow-up at eight. Read from the abstract, with staff ready to step in.
Limits: Compared with a waitlist, not with human therapy. Short, and run by its developers. It does not show the chatbot can operate unsupervised.
Checked against the original on November 6, 2026.
Open the source (opens in a new tab) - Established now
As of November 2025, FDA said no AI-enabled device was authorized for mental health uses.
What this does not prove: A status on that date, which may have changed.
U.S. Food and Drug Administration, November 6, 2025
Digital Health Advisory Committee: generative AI-enabled digital mental health devices
Scope: U.S. devices, as of the meeting date.
How it was done: FDA's executive summary for an advisory meeting.
Limits: A status as of that date, which may have changed. An August 2026 FDA discussion paper asks for input and is not guidance.
Checked against the original on November 6, 2026.
Open the source (opens in a new tab)
What may change
- Coaching, screening, long-term support and therapy-adjacent care.
What may endure
- Serious needs for connection, safety and agency.
The hard questions
- What should a human mental health professional uniquely provide?
- Could AI help underserved people while introducing other harms?
Evidence worth watching
Randomized outcomes · Harms · Regulation · Patient preference
Sources last checked November 6, 2026
Next review by February 6, 2027.
What changed, and how we decideCareer guidance, not medical, legal or financial advice. For a major loan or licensing decision, talk with a qualified professional, and bring your questions from here.
Do you disagree?
If you have counterevidence, or we got something wrong, we want to see it. Don't take our word for any of this.
Send us the evidence