Increasing AI Use in Medicine: How Will It Impact Care?

A new survey suggests that AI use is surging in the medical field, and doctors are beginning to embrace it as a tool in their practice. While it could be helpful for administrative tasks, its ability to assist with patient care has sparked concerns among doctors and patients.

Artificial intelligence (AI) has infiltrated almost every aspect of daily life. It's used in smartphones and smart home devices and as a tool to create written documents in record time. AI also has the potential to transform medicine, as it can read diagnostic tests, rapidly analyze data, and assist in diagnosing patients.

According to a new American Medical Association (AMA) survey, physicians are warming up to the idea of using AI in their practice. The Augmented Intelligence Research survey, first conducted in 2023 and followed up in 2024, showed that around 66% of the 1,183 doctors surveyed in 2024 said they currently use AI, up significantly from 38% in 2023.

Moreover, 68% of physicians surveyed in 2024 indicated they see advantages to using AI for patient care, a slight increase from 65% in 2023.

While promising as a physician's tool, the rise in AI use in the medical field has raised concerns about the potential benefits and risks for patients.

Survey highlights how doctors use AI

The 2024 AMA survey showed that just over half of physicians believe AI could be most beneficial for "addressing administrative burden through automation." In addition, 75% said it would help work efficiency, and 62% felt AI could improve clinical outcomes.

Moreover, 72% of respondents said AI's diagnostic ability would be helpful in clinical practice.

Doctors surveyed also said AI could be a tool for handling many aspects of healthcare.

These include:

• Translation services
• Creating discharge instructions, care plans, and progress notes
• Documenting billing codes, medical charts, or visit notes
• Developing chatbots for patient-related customer service functions
• Summarizing medical research and standards of care
• Assisting with diagnosis
• Generating chart summaries
• Creating health recommendations

However, the survey results showed that doctors are most enthusiastic about using AI to document billing codes, medical charts, or visit notes and create discharge instructions, care plans, and progress notes.

Generating responses to messages in patient portals is also on the respondents' list of top potential uses for AI.

Still, many physicians would like to see a designated feedback channel (88%), data privacy assurances (87%), and electronic health record integration (84%) integrated into AI tools to consider using them.

While nearly three-quarters of respondents said AI's diagnostic ability would be helpful in clinical practice, survey data found that in 2024, 51% either currently or immediately plan to use AI to assist with diagnosing patients.

Dr Jagdish Khubchandani, a professor of public health at New Mexico State University, tells WellnessPulse that globally, the use of AI in medicine is inevitable and will continue to grow for several reasons.

"Why did we start using calculators when we could compute basics (e.g., addition, subtraction, etc.)? Because of ease, convenience, utility, efficiency, and speed," Khubchandani explains. "So, we will see more and more providers and facilities using AI, just like in other domains of our society like [the] education sector."

Potential problems with using AI to diagnose health conditions

AI-assisted diagnosis involves algorithms that can quickly identify and diagnose health conditions by analyzing medical images like X-rays or MRIs, and patient data.

Recent research found that the average diagnostic accuracy for physicians using the AI platform Chat GPT Plus was better than that of doctors who used conventional diagnostic approaches. However, there was only a 3% difference between the accuracy of the two groups.

Khubchandani suggests that the rising use of AI might mean that healthcare providers see more benefits than harms of traditional and generative AI like Chat GPT. Still, there are concerns that must be addressed.

The major issues pertain to reliability, validity, accuracy, and issues related to providers and patients. The biggest issue, in my view, would be the first three as lives are at risk and […] AI can be unreliable at times given the nature of how AI works and is implemented.

Dr Jagdish Khubchandani

Quality healthcare depends on compassionate human interaction, which, Khubchandani says, cannot be fully substituted by machines.

"This can be labeled as lack of satisfaction in care and often, may lead to or coexist with mistrust and reduced time with the providers," he explains. "Other patient concerns may relate to privacy issues, misdiagnosis, and cost-related issues."

Regulatory issues with AI in medicine

The AMA survey showed that 47% of doctors believe there is a need for increased oversight of AI-enabled medical devices by governing bodies such as the U.S. Food and Drug Administration (FDA).

In a 2024 editorial published in the Journal of Medicine, Surgery, and Public Health, Khubchandani and his colleagues said that unless scholars worldwide produce robust scientific evidence on the value and impact of AI in healthcare, the establishment of regulatory or ethical frameworks and industry standards will remain an elusive goal.

"The regulation worldwide is in nascent stages, [as] I serve on the FDA DHAC as a consultant that met [for the] first time in Nov 2024 to discuss issues related to [the] regulation of generative AI," Khubchandani explains. "Technology grows and spreads faster than regulatory bodies, a lot is evolving, and this is a new domain for agencies."

Patients are concerned about AI

A 2023 Pew Research survey found that 60% of adults in the U.S. would feel uncomfortable if their healthcare provider relied on AI for diagnosing and treating their medical conditions. However, 38% believe AI use in the doctor's office would improve outcomes, while 33% think it would lead to worse outcomes. Around the same number of respondents said it wouldn't make a difference.

Overall, three-quarters of adults are concerned that the rise in physicians using AI to diagnose and recommend treatments is happening too fast, without much understanding of the potential risks for patients.

Dr Kwadwo Kyeremanteng, Head of ICU at The Ottawa Hospital in Ottawa, Ontario, Canada, tells WellnessPulse that from a research standpoint, AI enables more precise and efficient queries and can produce high-quality, cited references that enhance decision-making.

I've personally found AI useful in researching complex medical cases. However, one major concern is the data AI is trained on. There are potential biases, particularly regarding race-based data. Will AI-generated insights be applicable to diverse patient populations, including BIPOC communities, or will they be skewed toward a predominantly Anglo-Saxon male dataset? This is an essential consideration in ensuring equitable patient care.

Dr Kwadwo Kyeremanteng

Still, Kyeremanteng says that AI can introduce objectivity into the diagnostic process, reducing human biases that may be influenced by a clinician's experience and personal biases. It can also offer an additional perspective, which could help physicians avoid diagnostic blind spots.

"I currently use AI to assist with complicated diagnoses and research challenging cases. In the future, I plan to integrate AI more into patient encounters as reliable tools become available," Kyeremanteng says.

The outlook for AI in medicine

According to the AMA survey, healthcare providers are primarily interested in using AI to reduce administrative burdens, such as documenting patient visits or creating care plans. While its ability to assist doctors with diagnosing patients is appealing, the consensus among medical professionals is that AI still needs to evolve, with appropriate oversight in place before many will trust the technology.

"Keep in mind, AI should be used as a tool for support rather than as a replacement for clinical judgment," Kyeremanteng suggests. "No AI model is perfect, and over-reliance without proper oversight could lead to misdiagnoses or gaps in patient care."

In the end, since AI use in medicine is in its infancy, it's too soon to tell exactly how it might impact patients.

Nonetheless, moving forward, regulatory agencies can adopt a measured approach, carefully considering IA's potential impact on healthcare to help physicians and patients embrace the benefits of this technology while ensuring it's used responsibly.