Gen AI adoption in healthcare

Generative artificial intelligence is moving from pilot projects to daily use inside U.S. healthcare organizations. McKinsey said on April 16 that half of surveyed healthcare leaders had already implemented it. (mckinsey.com) The survey covered 150 U.S. leaders across payers, clinical-care organizations, and healthcare services and technology firms, and it was fielded from Sept. 17 to Oct. 17, 2025. More than 80% said their organizations had already deployed at least one use case to end users. (mckinsey.com) McKinsey said adoption rose from 25% in late 2023 to 47% in 2024 and 50% by the end of 2025. Half of respondents also said their organizations launched their first use cases more than six months earlier, a sign that some systems are now past the test phase. (beckershospitalreview.com) In healthcare, generative artificial intelligence usually means software that drafts text, summarizes records, answers questions, or helps workers search large piles of documents. Hospitals and insurers have aimed those tools first at paperwork-heavy jobs because billing, prior authorization, call centers, and clinical documentation generate huge volumes of text. (mckinsey.com) That helps explain where leaders say the technology is landing first. McKinsey found administrative efficiency had the biggest perceived upside, while clinical productivity was already among the most widely implemented use cases, with more than half of care-organization respondents reporting clinical-productivity deployments. (beckershospitalreview.com) The bottleneck has shifted as adoption rose. McKinsey said safety risks such as inaccuracy, bias, security, and regulatory compliance still rank high, but leaders now also cite integration with existing systems and a lack of internal capabilities as major barriers to wider rollout. (mckinsey.com) (beckershospitalreview.com) Federal regulators have already started pushing for more visibility into how some healthcare algorithms work. The Office of the National Coordinator for Health Information Technology said its HTI-1 final rule added transparency requirements for certain artificial intelligence and predictive algorithms in certified health information technology. (healthit.gov) Doctors are using these tools more often too, but the caution has not disappeared. The American Medical Association said in March 2026 that its physician survey found AI adoption continuing to grow, while earlier AMA results showed physicians still wanted stronger oversight, data privacy protections, and proof that tools work accurately in practice. (ama-assn.org 1) (ama-assn.org 2) McKinsey said 82% of surveyed leaders expect a positive return on investment, and 45% said they had already quantified returns, usually at less than two times to four times their initial investment. The next fight is less about whether to try generative artificial intelligence than whether healthcare organizations can wire it into real workflows without creating new risks. (beckershospitalreview.com)