Sonographers are not going to be replaced anytime soon. The U.S. Bureau of Labor Statistics projects 13% employment growth for diagnostic medical sonographers from 2024 to 2034, which is much faster than average. That means roughly 11,700 new jobs will be added, bringing total employment from about 90,000 to nearly 102,000. AI is changing what the job looks like, but the forces driving demand for sonographers, including an aging population and expanding use of ultrasound, are outpacing any labor-saving effects of new technology.
What AI Can Actually Do Right Now
AI has made real progress in ultrasound, but its capabilities are narrower than headlines suggest. Current systems can automatically identify standard imaging planes during fetal ultrasounds, flag when image quality is poor, and provide real-time feedback on probe positioning. Some tools run embedded processors that perform quality checks as images are captured, comparing what’s in the image against what should be there based on anatomical mapping.
One well-studied AI system called S-Detect, designed for thyroid nodule assessment, illustrates both the promise and the ceiling. In a retrospective study comparing it against sonographers of different experience levels, S-Detect achieved 98.4% sensitivity and 78.4% specificity when evaluated alongside junior sonographers, who scored 96.9% sensitivity but only 52.9% specificity. Against senior sonographers, the picture flipped: S-Detect matched their sensitivity (97.5% vs. 96.7%) but fell short on specificity (57.7% vs. 69.2%). In plain terms, AI is very good at catching potential problems but less accurate than experienced humans at distinguishing real threats from false alarms.
This pattern matters. High sensitivity with lower specificity means more unnecessary follow-up testing, more patient anxiety, and higher costs. Experienced sonographers bring contextual judgment that helps filter out noise, and that skill gap hasn’t been closed.
Why the Physical Job Is Hard to Automate
Ultrasound is fundamentally a hands-on process. A sonographer holds a transducer against a patient’s body, adjusts pressure and angle in real time, and responds to what they feel and see simultaneously. Robotic ultrasound systems exist in research settings, but probe control remains a core technical challenge. Planning a scanning path and executing it on a living, breathing person who shifts position, tenses muscles, or has unusual anatomy is far more complex than automating a static imaging task like analyzing an X-ray that’s already been taken.
Beyond the physical side, sonographers exercise clinical judgment throughout every exam. They adapt their technique when a patient’s body type limits image quality. They recognize when findings fall outside their scope and need to be escalated. They decide in real time whether an image is diagnostic or needs to be recaptured from a different angle. These decisions draw on pattern recognition, anatomical knowledge, and years of clinical experience. AI guidance tools can assist with some of these decisions, but they can’t replace the integrated thinking a sonographer performs while simultaneously operating the equipment.
The Liability Problem No One Has Solved
Even if AI reached human-level diagnostic performance tomorrow, a major legal barrier would remain: nobody has figured out who’s responsible when an AI-assisted ultrasound gets it wrong. Current malpractice law assumes a human made a decision using professional reasoning. When that reasoning is replaced or heavily influenced by an algorithm, courts struggle to assess whether the diagnosis met accepted standards.
Many AI diagnostic models operate as “black boxes,” meaning their internal logic can’t be easily examined or explained. This creates problems on multiple fronts. Clinicians may develop automation bias, trusting the AI output without fully evaluating it themselves. And when errors occur, responsibility gets distributed across the clinician, the hospital, and the software developer in ways existing legal frameworks weren’t built to handle. Until these liability questions are resolved through regulation or case law, healthcare systems have strong incentives to keep trained humans at the center of the diagnostic process.
How POCUS Is Reshaping the Field
Point-of-care ultrasound, or POCUS, lets physicians and nurses perform focused scans at the bedside without sending patients to an imaging department. This trend is growing rapidly, and AI-enabled portable devices are making it easier for non-specialists to capture usable images. In theory, this could reduce demand for sonographers in certain settings.
The reality is more nuanced. POCUS exams are typically limited in scope, answering a specific clinical question (like whether there’s fluid around the heart) rather than performing a comprehensive diagnostic study. When a POCUS scan reveals something unexpected or complex, patients still get referred for a full exam performed by a sonographer. AI tools that help non-specialists capture basic images may actually increase the volume of follow-up studies that sonographers perform. Some professionals in lower-income countries have raised concerns about displacement, noting that if “everybody and anybody can scan” with AI assistance, it could create professional conflicts. But in settings with existing sonographer shortages, the more immediate effect is expanding access rather than eliminating jobs.
What the Job Will Look Like
The most realistic near-term future is one where AI handles repetitive, time-consuming parts of the workflow while sonographers focus on complex cases and clinical decision-making. AI can pre-screen image quality, auto-label anatomical structures, and flag abnormalities for review. This could reduce the tedious elements of the job and potentially help with the high rates of repetitive strain injuries that sonographers experience from long hours of scanning.
Sonographers who develop skills in working alongside AI tools, interpreting their outputs critically, and understanding their limitations will be best positioned. The profession is also expanding into education. As POCUS spreads across medical specialties, experienced sonographers are increasingly needed to teach other healthcare professionals how to scan properly. Digital simulators and AI-based training tools are useful for building foundational skills, but research confirms they don’t replace the need for learning in a clinical environment with experienced educators providing feedback and assessing competency.
The 13% growth projection through 2034 reflects all of these dynamics: aging populations needing more imaging, ultrasound being used in more clinical settings, and the technology creating new roles even as it changes existing ones. For anyone considering sonography as a career, the data points toward a field that’s evolving, not shrinking.