AI Shouldn’t Be Your MD

AI Shouldn’t Be Your MD

By Amelia Nugent, Tulane University MPH in Health Policy student & Louisiana Progress Graduate Student Intern

When I was 17, I had to get my tonsils removed. I was getting strep throat way more often than normal, and the ENT thought that would be the best course of action to prevent recurrence. My health insurance covered the procedure because they deemed it medically necessary, and I have not gotten strep since. 

However, a tonsillectomy at the age of 17 is unusual, as the procedure is usually recommended for younger children. I had not had issues with strep when I was a child, so there had been no reason to remove my tonsils at the normal age. Had AI reviewed this situation, I may have been denied the surgery because of my age and given that I was otherwise healthy, because that is the only information that would have been provided to the algorithm. This would have left me with two choices: continue to get strep every few months, leading to me having to stay home from school and miss educational opportunities, or pay thousands of dollars out of pocket for the procedure. Luckily, I did not have to face this decision, but it is a choice that many are having to make as a result of insurance companies using artificial intelligence to deny care that is medically necessary. 

As advanced as AI is becoming, it does not have the situational awareness and individualized medical experience of human physicians, 60% of whom are concerned that health insurer’s use of AI is increasing prior authorization denials. This issue has led to a class action lawsuit alleging that the AI models used by major insurers to review claims have an error rate of up to 90%. Patients are worried as well. Approximately 70% of patients say they are concerned that AI will not be fair in adjudicating claims, and about 50% stating that they trust a human claims reviewer more than AI. 

Other states across the country are acknowledging that this is an issue, too, with legislation regulating AI use by health insurers enacted or pending in 25 states and Louisiana, as of March 3, 2026, including Texas, Alabama, Tennessee, and Oklahoma. SB246 aligns our AI policy with these other Southern states, ensuring that Louisiana does not get left behind in terms of consumer protection and patient safety. 

The law acknowledges that AI can be useful for health insurers in terms of operational efficiency, but not at the expense of the health of premium-paying Louisianans. As such, SB246 requires that AI does not replace the role of a human in the claims determination process, and that it cannot be used to deny a health insurance claim or prior authorization request that is based on medical necessity without review from a medical professional. Health insurers must also disclose the use of AI in their review process. These regulations represent a positive step forward for patient protection without compromising the benefits that AI provides to health insurers.