As artificial intelligence continues to raise concerns about job losses, a new analysis by IT support provider Hamilton Group suggests that some professions remain difficult to replace with AI, particularly jobs that depend heavily on human judgment, physical presence and direct interaction.
Emergency medical technicians, paramedics and several healthcare and social service professionals are among the occupations identified by Hamilton Group as having little to no AI automation risk.
The September 2026 report examined 60 occupations with an AI automation risk below 45% and compared their responsibilities with AI augmentation data from the Anthropic Economic Index, along with current employment and median wage figures.
The findings come as AI adoption continues to spread across workplaces. While some tasks can increasingly be handled by AI, Hamilton Group’s analysis suggests that the technology’s impact may look different in professions where workers must make decisions in real-world situations and interact directly with other people.
Emergency medical technicians had a 0% AI automation risk in the Hamilton Group analysis. The occupation employed about 181,000 people in the U.S. in 2025, with a median annual wage of $44,000, according to the report.
EMTs work in situations where decisions have to be made quickly and often under unpredictable conditions. They assess patients, provide emergency care and respond to 911 calls, making physical presence and real-time judgment central to the job.
READ: ‘Is that the meaning of life?’: Bernie Sanders questions tech CEOs on AI future for workers (May 29, 2026)
Paramedics also had a 0% automation risk in Hamilton Group’s analysis. About 101,000 people were employed in the occupation in 2025, with a median wage of $61,000.
While AI could potentially assist with administrative work, such as preparing reports after a shift, the core responsibilities of a paramedic involve assessing patients and providing care in emergency situations.
The Hamilton Group report also found relatively low automation risks across several physician and social service occupations.
Physicians classified as “all other” had an estimated 4.5% AI automation risk, while their median annual wage was $266,000, the highest among the 10 occupations highlighted in the analysis.
Mental health and substance abuse social workers also had a 4.5% automation risk. Their work often involves direct conversations with people dealing with complex personal circumstances, where understanding context and building trust are important parts of the job.
Child, family and school social workers had a 6.7% automation risk. The occupation employed about 393,000 people in 2025, making it the largest profession among the 10 occupations identified by Hamilton Group.
Registered nurses had a 6.9% automation risk, with about 3.4 million people employed in the profession and a median annual wage of $98,000.
Overall, the 10 occupations highlighted by Hamilton Group have AI automation risks ranging from 0% to 10.3%. EMTs and paramedics have the lowest risk at 0%, followed by physicians and mental health and substance abuse social workers at 4.5%. Child, family and school social workers have a 6.7% risk, while registered nurses stand at 6.9%. Clinical and counseling psychologists, nurse practitioners and general internal medicine physicians have risks below 10%, at 9.8%, 9.9% and 9.9%, respectively. Coaches and scouts have the highest risk among the 10 at 10.3%.
Together, these professions represent millions of U.S. workers, with registered nurses accounting for the largest workforce at 3.4 million. Median wages across the 10 occupations range from $44,000 for EMTs to $266,000 for physicians.
Healthcare and social services account for eight of the 10 occupations identified by Hamilton Group, with each recording an automation risk below 10%.
The report also highlights the difference between AI automation and AI augmentation. A profession can incorporate AI into its work without the technology replacing the person performing the job.
That distinction is particularly relevant to healthcare. AI tools can assist medical professionals with tasks such as analyzing information, organizing records and accessing research, while trained professionals remain responsible for patient care and decisions.
READ: AI adoption linked to 13% job loss among young U.S. workers, Stanford finds (August 28, 2025)
Hamilton Group’s analysis points to a broader trend: the professions with the lowest estimated replacement risk are often those where the job involves more than producing information.
A doctor still has to interact with a patient. A nurse still has to provide hands-on care. An EMT has to respond to an emergency. A social worker has to understand a person’s circumstances and work with them directly.
“The human factor is a major part of the job of a doctor, a nurse, a social worker, or a paramedic,” a Hamilton Group job analyst said in the report. “Their ability to establish connection, understand the problem, and extend empathy to another person is what AI is unable to do.”
The Hamilton Group analyst also pointed to a less-discussed aspect of the findings: many of the jobs with low AI automation risk are not among the highest-paid professions.
Among the five occupations with the lowest automation risk, physicians were the only ones with a median annual wage above $61,000.
The findings do not mean these professions will remain completely untouched by AI. Instead, Hamilton Group’s analysis suggests that AI could assist with specific tasks while humans continue to perform the parts of the job that require physical action, judgment, communication and responsibility.
As AI becomes more capable, that distinction may become increasingly important. For many workers, the question may not simply be whether AI can do their job, but which parts of their job AI can actually do and which parts still require a person.