AI tools in healthcare may add value in novel and non-traditional ways.
getty
Over the past two years, there has been notable progress in the implementation of AI in healthcare. Organizations are now considering moving from small pilot projects to expansive AI transformation programs. The critical question remains: have these tools demonstrated enough value to warrant further investment, and how should their impact in healthcare be assessed?
The question of investment and value capture varies widely across the healthcare sector. In July, UnitedHealth Group, a leading payor organization, announced an investment of nearly $1.5 billion in AI across various units, including insurance, back-office automation, care delivery, and technology operations. Competitors like Humana and Centene are making similar investments, anticipating long-term productivity and operational success.
Regarding return on investment (ROI), there is emerging evidence that substantial value can be realized if AI tools are effectively integrated into existing systems. A 2026 Productive/Edge report suggests that for every dollar invested in AI, a return of approximately $3.20 is achieved over 14 months. Additionally, the study found that organizations could achieve nearly 147% ROI within three years, with 45% seeing positive returns within the first year.
However, not all studies paint a positive picture. Earlier reports often highlighted the failure of AI investments to deliver returns. A pivotal study by MIT found that nearly 95% of AI pilots did not achieve measurable returns, causing concern across the industry. Similarly, a McKinsey study reported that while a significant number of companies have invested in AI, many see no substantial impact on their financial outcomes.
The ambiguity in ROI highlights a fundamental issue with adopting innovative technology: conventional methods may not effectively measure ROI. The differing opinions on AI’s value reflect varying interpretations of how value is captured. Consequently, the definition of “value” needs to be clearly articulated.
Consider ambient scribing as an example. Research is increasingly revealing benefits to this technology. A recent JAMA study indicated that AI scribe adoption reduced EHR time by 13.4 minutes, documentation time by 16.0 minutes, and allowed for 0.49 additional weekly visits. Though these time savings might seem minimal, they accumulate across multiple physicians and clinics. Another study showed that AI in radiology reduced interpretation times for brain CT lesions by 11.23%, lung lesions by 52.82%, and blood smear analysis by 61%.
Although these figures might not lead to substantial cost savings or increased patient visits, they are crucial. They suggest that ROI should be evaluated with specific considerations, such as physician satisfaction. Saving a few minutes per patient encounter can significantly enhance a physician’s quality of life, which is vital given the high rates of burnout in healthcare. Moreover, as AI tools become integral to medical education and practice, systems that fail to adopt them might struggle to attract and retain talent.
It is well documented that healthcare margins are razor-thin. The median operating margin for U.S. health systems was -0.1% in 2023, improved to 1.6% in 2024, dropped to 1.0% in 2025, and stood at 0.4% as of March 2026. Consequently, organizations must ensure tangible ROI before investing billions in AI. However, as seen with past disruptive technologies, traditional value models may not align with modern advancements. AI represents a significant technological shift, particularly in medicine, necessitating a reevaluation of outdated financial strategies to define “value” in today’s context.
