The Healthcare AI Alignment Problem: Key Gaps Between Patients and Organizations by Paige Nong, PhD

The Healthcare AI Alignment Problem: Key Gaps Between Patients and Organizations by Paige Nong, PhD

Title history
  • The Healthcare AI Alignment Problem: Key Gaps Between Patients and Organizations by Paige Nong, PhD — observed on 20261030-1104 UTC (YouTube)

by Paige Nong, PhD · View the channel on PSI

Humanities, Social Sciences & Thought AI & Society UBDigital and Information technology: general topicsUBJDigital and information technologies: social and ethical aspects
🎙 Paige Nong, PhD 👥 1K 📅 October 28, 2026 ⏱ 58 min 👁 49 📄 original study 🧭 2026-10-30
Available in: English (current) Français

Keywords

AI alignmenthealthcare deliverypatient perspectivesorganizational governancenational surveys

Summary

Paige Nong, a professor at the University of Minnesota, presents a seminar on the alignment problem in healthcare AI. She defines two main categories of AI: predictive (used for decades) and generative (recent). Using data from the American Hospital Association’s IT supplements, she shows a rapid increase in AI adoption: 66% of hospitals used predictive AI in 2023, but only 44% evaluated it regularly. By 2025, 77% of hospitals had adopted generative AI, with ambient scribes being the most common application. However, evaluation and governance lag behind implementation. Qualitative interviews across various hospitals reveal wide variation in governance processes, from robust committee-based approaches to ad-hoc decisions. Patient engagement also varies: academic medical centers rarely include patients in AI governance, while federally qualified health centers do by design. The talk highlights a critical gap: patients are often unaware of AI use, and there is no federal requirement for disclosure. This creates risks to patient trust. The speaker discusses implications for ethical alignment and calls for better standards and patient involvement.

169 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable empirical evidence on the state of AI in US healthcare, combining national quantitative data with in-depth qualitative insights. The argumentation is solid: the speaker systematically builds a case that implementation outpaces governance and patient awareness, using concrete data points on adoption rates and evaluation gaps. She effectively contrasts predictive and generative AI to highlight the unprecedented speed of adoption. The qualitative interviews add depth, showing real-world governance challenges. The discussion of patient perspectives and legal cases in California strengthens the argument about missing disclosure requirements. Overall, the presentation is well-structured, evidence-based, and makes a compelling case for the pressing need to address AI alignment in healthcare.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high. The speaker cites specific national surveys (AHA IT supplement) conducted in collaboration with the ONC, and mentions her research published in Health Affairs and JAMA Network. She discloses funding sources and declares no conflicts of interest. The methodology is transparent, with survey data and qualitative interviews described clearly. The title precisely matches the content, focusing on the alignment gap between organizations and patients. The talk stays on topic throughout, never straying into unrelated material. While no direct external citations with URLs are provided in the description, the references to reputable institutions and peer-reviewed outlets support the reliability. No comments were available for analysis.

233 words

Title / Content Match

The title accurately reflects the content: it focuses on the gap between how healthcare organizations use AI and what patients expect, highlighting an alignment problem. The talk delivers on this promise by presenting empirical evidence on both sides.

Quality & Reliability

9/10

The talk presents rigorous empirical data from national surveys and qualitative interviews, conducted by a researcher at a reputable institution, with findings published in major journals. The methodology is transparent and the speaker discloses funding sources and conflicts of interest.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • American Hospital Association Annual Survey IT Supplement — Used to collect nationally representative data on hospital AI use and evaluation
  • Office of the National Coordinator for Health Information Technology (ONC) — Collaboration on survey design and analysis of AI adoption trends
  • Health Affairs — Journal where the speaker's research on AI governance and patient perspectives is published
  • JAMA Network — Another journal publishing the speaker's work on health IT implications

Contribution & Novelties

This talk contributes original empirical data on the rapid adoption of generative AI in US hospitals, revealing that 77% of hospitals used it by 2025—far exceeding earlier projections. It also highlights significant gaps in evaluation and governance, particularly for generative AI, and underscores the lack of patient representation in AI governance processes. The speaker’s qualitative interviews with rural and academic hospitals provide novel insights into the variability of governance practices. The discussion of disclosure and legal issues adds a practical dimension. Overall, it advances the understanding of the AI alignment problem in healthcare by bridging organizational and patient perspectives.

Pour aller plus loin :

158 words

Radar Profile

The radar profile shows strong scores in information quality and reliability, with slightly lower but still high scores in quantity and technical level. This indicates a well-researched, evidence-driven presentation that balances depth with accessibility, making it both informative and trustworthy.

Reliability 9/10