Key Takeaways
- America’s major hospital systems are increasingly adopting artificial intelligence (AI) for various tasks, with 27% of health systems paying for commercial AI licenses.
- AI is being used for labor-intensive tasks such as dealing with insurance claims, taking notes, and reviewing X-ray reports.
- While AI has shown benefits, such as speeding up the review of scans, it also has limitations and can make errors, highlighting the need for human oversight.
- Doctors are adopting a "trust but verify" approach when using AI tools, recognizing the need to verify the accuracy of information provided by AI systems.
Introduction to AI Adoption in Healthcare
America’s major hospital systems are becoming a testing ground for widespread artificial intelligence (AI) adoption, according to a report by the Wall Street Journal (WSJ). The report notes that these health centers are seeing what AI can do, but also showing where it falls short, in some cases through disturbing errors. As the WSJ reports, "A little more than a quarter of health systems (27%) are now paying for commercial AI licenses, a figure that’s triple the rate across the U.S. economy." This significant adoption rate highlights the potential of AI to transform the healthcare industry.
AI Use in Labor-Intensive Tasks
AI use is particularly prominent for labor-intensive tasks that hospitals deal with each day, such as dealing with insurance claims or taking notes. For example, Samir Abboud, chief of emergency radiology for Northwestern Medicine, uses AI to review X-ray reports. As Abboud notes, "You’d feel guilty getting up to use the restroom. There’s hundreds of patients waiting for our read, and any one of them could be one that’s actively dying." This quote highlights the pressure on healthcare professionals to review scans quickly and accurately, and how AI can help alleviate some of this pressure. By using AI to review X-ray reports, Abboud and his team can focus on more complex cases and provide faster care to patients.
Limitations of AI in Healthcare
However, there are limits to what AI can do for doctors, as the WSJ reports. For instance, Mayo Clinic cardiologist Paul A. Friedman used ChatGPT to get more information on a defibrillator implantation, but the chatbot provided fabricated references to medical journals. This experience led Friedman to adopt a "trust but verify" approach, stating, "It’s not that I don’t ask ChatGPT medical questions but, when I do, I always look for the references, click on them and read the abstracts at a minimum." This approach highlights the need for human oversight and verification of AI-provided information to ensure accuracy and safety.
Current State of AI Adoption in Healthcare
Research shows that close to half of healthcare and life-sciences organizations have generative AI in production use, in many cases for documentation, administrative work, and early-stage clinical summaries. Additionally, upwards of half of doctors surveyed by the American Medical Association said that AI tools could meaningfully support core clinical functions. As reported, "Among respondents, 72% said AI could improve diagnostic ability, 62% said it could enhance clinical outcomes and 59% said it could strengthen care coordination." This data suggests that AI has the potential to make a significant impact on the healthcare industry, but it is crucial to address the limitations and challenges associated with its adoption.
Conclusion and Future Directions
In conclusion, America’s major hospital systems are at the forefront of AI adoption, with 27% of health systems paying for commercial AI licenses. While AI has shown benefits, such as speeding up the review of scans, it also has limitations and can make errors. As the healthcare industry continues to adopt AI, it is essential to recognize the need for human oversight and verification of AI-provided information. By adopting a "trust but verify" approach, healthcare professionals can ensure that AI is used safely and effectively to improve patient care. As the industry moves forward, it will be crucial to address the challenges associated with AI adoption and to develop strategies for mitigating errors and ensuring the accuracy of AI-provided information.