
The precipice of real AI-driven change in health care: What it means for outcomes and risk
Key Takeaways
- Medical malpractice severity is escalating, with nuclear verdict frequency and award size increasing, driving sustained premium inflation and greater hospital risk retention.
- AI can introduce new failure modes and accountability ambiguity, so safety assessments must focus on measured outcomes rather than tool count or theoretical benefit.
If artificial intelligence improves medicine, what does that mean for malpractice insurance premiums?
Artificial intelligence (AI) has penetrated the health care industry as it has almost every other sector — and, as in almost every industry it touches, in health care, fundamental questions of adoption remain unanswered. Hospital operators, physicians and insurers are all asking the same thing: What does more AI implementation mean for patient safety and institutional risk? The honest answer is that we don't fully know yet. A lack of data and historical evidence to effectively determine what kinds of AI tools, and AI use, really impact efficiency and positive outcomes has made adoption more complex.
Meanwhile, the stakes have never been higher. Data show that there was a 52% increase in
The opportunity and excitement around AI in health care persist, rightly so, but hospitals and physicians need a clearer path for risk analysis and safety.
The seduction of more
There's an intuitive appeal to more AI. More diagnostic support, better documentation, the list goes on. Theoretically, these should reduce errors and improve safety. We believe targeted AI applications will improve safety for doctors and patients alike. But as more tools enter the market, one thing is clear: More tools don't always guarantee safer care. Tools can also create new failure modes. Decision-making can shift in ways we don't fully understand yet: Cognitive load changes, and accountability becomes murky. That's why measuring outcomes is the critical component to determining how AI is impacting our health care systems. Specific tools may drive better judgment or outcomes than others.
If doctors become less error-prone as they implement AI, we will eventually see that play out in overall outcomes. We're in a period of serious evolution for hospitals and physicians as they onboard new tools, understand what works and assess where AI implementation makes sense. The opportunity is real, but the path forward requires intellectual honesty about what we actually know and what we're still guessing at.
What's actually working
So where are we right now? Evidence of positive outcomes using AI is strongest in the following areas:
- Enhanced perception: diagnostic imaging tools that catch what human eyes might miss (for example, in radiology and pathology)
- Reduced cognitive burden: administrative tools that free mental bandwidth for clinical judgment
- Standardized protocols: AI-driven pathways that eliminate variation in care decisions
- Real-time feedback: tools that flag potential errors before they harm patients
What we're learning is that the best outcomes come when practitioners understand why a given tool flagged something — not when they just follow its recommendation — and receive support in next steps. This changes how we should think about training, implementation and risk associated with AI-assisted diagnoses.
Pricing follows the evidence
What we know is that health care isn't reinventing the wheel when it comes to AI implementation and safety. The industry will follow a path other insurance verticals have already walked. In homeowners insurance, for example, smart smoke alarms, water-leak sensors and connected security systems produced measurable loss reductions — and carriers passed those savings back as premium credits. In auto, the same dynamic is now playing out with driver-assist and autonomous features: Lemonade, for instance, discounts premiums for Tesla drivers using full self-driving mode.
The pattern is consistent: Sensors and software generate outcome data, and pricing follows the evidence. Medical malpractice will move the same way, but only once AI tools produce the equivalent evidence at scale. In the meantime, medical malpractice tools that take a precise approach will be the ones to help usher in this next era of AI-supported medicine. Hospitals and practitioners have already experienced
The precipice of change
I do believe that the numbers of errors and omissions will go down over time due to AI implementation, but we're not there yet. Over the next two to three years, we'll see a handful of AI solutions providers emerge with genuinely compelling data at scale to authoritatively prove the effectiveness of their tools. Hospitals and physicians will adopt those tools because the evidence is clear and the financial incentives are aligned. Premiums will begin to reflect actual risk rather than uncertainty. And medicine will move into a new era in which AI-supported care is the baseline, not the exception.
Jared Kaplan is CEO and cofounder of
Related to this article









