AI experts warn of ‘zombie algorithm phenomenon’ impacting imaging
Artificial intelligence experts are warning of a “zombie algorithm phenomenon” that affects medical imaging and other areas of medicine.
Scientists with Columbia University shared their concerns in a new opinion piece, published Feb. 4 in Health Affairs. They noted that tech giants are pouring $250 billion each year into AI and see the $5 trillion U.S. healthcare industry as a “primary target” to seek returns on their investment.
Yet, numbers “tell a troubling story.” One survey of U.S. healthcare executives conducted by Menlo Ventures shows the adoption of AI tools outpaces other industries by a factor of 2.2, reaching $1.4 billion last year alone. The MIT NANDA Initiative also estimates that that 95% of enterprise gen AI pilots fail to deliver measurable financial returns.
“This disconnect between promise and performance has created what economists increasingly recognize as a classic bubble,” Maxim Topaz, PhD, RN—whose research focuses on the application of AI and data science to improve healthcare quality and outcomes—and colleagues wrote Feb. 4. “Coverage of the 2024 JAMA AI Summit characterized medicine’s AI adoption as ‘flying blind,’ not as hyperbole but as a reflection of deployment outpacing evidence, oversight and evaluation,” they added later.
Topaz and co-authors gave three potential reasons why they believe the AI bubble “must burst.” They contend this coming course correction is "not a catastrophe to be avoided,” but instead a “chemotherapy for metastasizing the problem.” Experts believe these issues make the bubble’s continuation “more dangerous than its burst”:
1. The zombie algorithm phenomenon: “A diagnostic algorithm trained on 2024 data will progressively fail as disease patterns, imaging technology, and patient populations evolve. Yet hospitals remain contractually obligated and operationally dependent on these deteriorating systems, bearing full liability for any resulting errors.”
2. The innovation winter: “With AI capturing 54% of digital health venture funding in 2025, other crucial health care innovations are being starved…This capital monoculture means health systems are forced to adopt over-engineered AI solutions for problems that might be better solved with simpler technologies.”
3. The evidence vacuum: “Traditional medical innovation follows a deliberate path: research, trials, peer review, regulatory approval, and careful implementation. AI has inverted this process.”
Columbia experts are advising health systems to learn from past bubbles and take immediate action to prepare for what comes next. Possible responses can include implementing vendor stress testing, establishing internal AI governance, and preparing operational contingencies.
“When the AI bubble bursts (economic indicators suggest this is when, not if), healthcare will face significant disruption,” the authors advised. “Yet this correction also presents an opportunity,” they added later. “Freed from the pressure to adopt every AI solution or be labeled ‘behind,’ health systems can return to their core mission: improving patient outcomes through evidence-based practice. The AI tools that survive will be those that deserve to: products that solve real problems, backed by real evidence, from companies with real business models.”
Read much more in Health Affairs.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.