Imagine being sick in your doctor's office – and rather than turning pages of your medical history or running tests that take several days, your doctor can quickly put together data from your health records, genetic profiles, and wearable devices to help you decipher what's wrong.
This kind of rapid diagnosis is one of the great promises of artificial intelligence for use in healthcare. Advocates of technology say that over the next decades, AI could save hundreds of thousands, and even millions of lives.
Additionally, a 2023 survey found that if the healthcare industry significantly increases AI use, it could save up to US$360 billion a year.
However, artificial intelligence is almost ubiquitous, from smartphones to chatbots to self-driving cars, but its impact on healthcare has been relatively low up to date.
A 2024 American Medical Association survey found that 66% of US physicians had increased their AI tools to some extent from 38% in 2023. Additionally, 43% of US healthcare organizations added or expanded the use of AI in 2024, but many implementations remain exploratory, especially when it comes to medical decisions and diagnosis.
I am a professor and researcher studying AI and healthcare analytics. It attempts to explain why AI growth is gradual and how technical limitations and ethical concerns can hinder widespread adoption of AI by the healthcare industry.
Inaccurate diagnosis, racial bias
Artificial intelligence is great at finding patterns in large amounts of data sets. In medicine, these patterns can indicate early signs of a disease that human physicians may overlook. Or, they show you the best treatment options based on how similar symptoms or background responded. Ultimately, this leads to faster, more accurate diagnosis and more personalized care.
AI can also help hospitals run more efficiently by analyzing workflows, predicting personnel needs, and ensuring valuable resources such as operating rooms are most effectively used. By streamlining tasks that require hours of human effort, AI can enable healthcare professionals to focus more on direct patient care.
But because of all its power, AI can make mistakes. These systems are trained on data from real patients, but can be struggling if something unusual happens or the data does not match the patient's previous patients.
As a result, AI does not always provide accurate diagnosis. This problem is called algorithm drift – when an AI system works well in a controlled configuration, but loses accuracy in real situations.
Racial and ethnic bias is another matter. If the data contains bias, AI can provide inaccurate recommendations and lead to misdiagnosis, as it does not include sufficient patients from a particular racial or ethnic group. Some evidence suggests that this is already happening.
Data sharing concerns, unrealistic expectations
Healthcare systems are labyrinth in their complexity. The prospect of integrating artificial intelligence into existing workflows is difficult. Introducing new technologies like AI can disrupt everyday life. Staff need additional training to effectively use AI tools. Many hospitals, clinics, and doctor offices do not have the time, personnel, money or will to implement AI.
Also, many state-of-the-art AI systems operate as opaque “black boxes.” They unlock recommendations, but even developers may have a hard time explaining how to do so. This opacity conflicts with medical needs where decisions require justification.
However, developers are often reluctant to disclose their own algorithms or data sources to protect their intellectual property and because complexity can be difficult to distill. The lack of transparency cultivates skepticism among practitioners, which slows regulatory approval and erodes trust in AI output. Many experts argue that transparency is not merely an ethical importance, but a practical need for adoption in a healthcare setting.
There are also privacy concerns. Data sharing can threaten patient confidentiality. Medical AI systems require a vast amount of patient data to train and predict algorithms. If not properly processed, AI may disclose sensitive health information through data breaches or unintended use of patient records.
For example, clinicians who use cloud-based AI assistants to draft notes must ensure that unauthorized parties do not have access to their patient's data. US regulations such as the HIPAA Act impose strict regulations on health data sharing. This means that AI developers need robust protection measures.
Privacy concerns also extend to patient trust. People may refuse or even reject AI-guided care when they fear that their medical data will be misused by algorithms.
The grand promise of AI is a formidable barrier in itself. The expectations are enormous. AI is often portrayed as a magical solution that can diagnose any disease and revolutionize the healthcare industry overnight. Such unrealistic assumptions often lead to disappointment. AI cannot fulfill its promise immediately.
Finally, it takes a lot of trial and error to develop a well-functioning AI system. AI systems must undergo rigorous testing to ensure they are safe and effective. This can take years and even after the system is approved, adjustments may be required as new types of data and real-world situations will be encountered.
Incremental change
Today, hospitals are rapidly hiring AI scribes who listen during patient visits and automatically draft clinical notes, reducing paperwork and allowing doctors to spend more time with patients. Research shows that over 20% of doctors use AI to describe a summary of progress or discharge. AI is also becoming a quiet force for administrative work. Hospitals deploy AI chatbots to translate appointment scheduling, triage common patient questions, and language in real time.
Clinical use of AI is present, but is more limited. In some hospitals, AI is the second eye for radiologists looking for early signs of the disease. However, doctors are still reluctant to decide to hand it over to the machine. Currently, only about 12% rely on AI for diagnostic help.
It is enough to say that the transition to AI in healthcare is progressive. Emerging technologies need time to mature, and the short-term needs of healthcare still outweigh the long-term benefits. In the meantime, AI is likely to deal with millions and save trillions.
Turgay Ayer owns a stake in Value Analytics Labs, a healthcare technology company. He received funding from government agencies, including the NSF, NIH and CDC.
/Commentary of the conversation. This material of the Organization of Origin/Author is a point-in-time nature and may be edited for clarity, style and length. Mirage.news does not take any institutional position or aspect, and all views, positions and conclusions expressed here are the authors alone.
