 | ● |  |
|
|
|
| | | AI news you ought to know about: - Gen Xers and their elders tend to believe AI will do more harm than good. More than half of 2,900 or so American adults 50 and older place themselves in that somewhat cynical category, according to research pollsters at the University of Michigan’s Institute for Healthcare Policy and Innovation. Only a fraction of the field, 4%, have “a lot of” trust in AI-generated information, and 47% have “little to no” trust in it. On the bright side for AI proponents, the biggest single swath—if only by a whisker, 49%—have “some” trust in AI outputs. The trusters tend to have used AI on their own. More:
- A surprisingly low 14% report using AI to get health-related information. Among those who did so, some 47% said a human interaction in person or by phone would be better for that purpose.
- Meanwhile, half these older adults say they’re not confident in their ability to sniff out incorrect info in AI outputs. They’re especially wary of AI hallucinations and deepfakes. The latter include audio as well as video, an important data point since scammers increasingly use AI voice technology to swindle older folks by phone. (The survey drew responses from individuals as old as 97.)
- “Our new data show that people who report their physical or mental health as fair or poor were more likely to lack confidence in their ability to spot incorrect AI-generated information—and were less likely to trust AI-generated health information—than those in good to excellent health,” comments poll director Jeffrey Kullgren, MD, MPH, a primary care physician at the VA Ann Arbor Healthcare System and associate professor of internal medicine at the University of Michigan Medical School. “Clinicians, aging-focused organizations and healthcare organizations should all take notice.”
- More findings and analysis are here.
- Worse yet, patients of all ages tend to look askance at doctors who use AI. That’s according to researchers at the University of Wuerzburg in Germany who surveyed American adults in January. Publishing their findings July 17 in JAMA Network Open, the team reports that, in every scenario described in various hypothetical situations, respondents perceived an imaginary AI-using physician as significantly less competent, trustworthy and empathetic than physicians not using AI. What’s more, participating healthcare consumers indicated a “significantly lower willingness to make an appointment with the portrayed physician if any type of AI use was mentioned.” Research brief here.
- Clinicians are excited about AI, but enthusiasm is running well ahead of adoption. That’s not only in the U.S. but around the world. Elsevier’s 2025 “Clinician of the Future” report, released July 15, shows 41% of global healthcare workers believe that clinicians who use AI will deliver better care than those who don’t. Yet only 16% are presently using AI tools to help make clinical decisions. The publisher heard back from more than 2,200 doctors, nurses and other frontline caregivers representing some 109 countries. The researchers also found the ratio of clinicians using AI at work almost doubled over last year, spiking to 48% from 26%. And get this: The use of AI for work purposes is highest in China (71%) and lowest in the U.S. (36%) and U.K. (34%). Also of keen interest, 97% of AI-using clinicians report using a generalist tool such as ChatGPT while only 76% have used an AI tool developed for a specific clinical use case. Download the full report or an executive summary of it here.
- When Texas became the newest state to enact a comprehensive law to govern AI use, it probably set a precedent. And that’s just because of its sheer size and massive population. Gov. Gret Abbott signed the bill June 22. This week legal experts comment on what the development means for Texans and may come to mean for residents of other states. Called Traiga for the Texas Responsible Artificial Intelligence Governance Act, the new law “brings forth a new approach to AI regulation, both by limiting Texas’s ability to punish companies with prohibitions on only a few intentional harms and by expanding the state’s investigatory powers,” explain attorneys at Boston-based WilmerHale in a July 21 blog post. “When the law goes into force on January 1, 2026, the net effect will likely subject many private enterprises to regulatory review and few to actual punishment. Given the size and economic heft of Texas, the law will likely have substantial ramifications for the AI industry.” Read the rest.
- No one will benefit if rural healthcare systems fall through the cracks in the AI boardwalk. A healthcare expert with BDO’s U.S. operation warns about the potential in a video interview clip posted by Medical Economics. Rural health centers and community practices can avoid such a lousy outcome, says Brad Boyd, MBA, by clinically aligning with larger systems to leverage the latter’s EHR systems. “If they’re involved in those types of relationships, [resource-strapped] rural healthcare facilities can benefit from the bigger system’s ecosystem of IT tools, which could include AI,” Boyd explains. “If you’re really independent, all on your own, you’re somewhat limited. You don’t have the capital to make investments like a world-class academic medical center.” It would be unfortunate for all stakeholders—not least the affected residents—if rural provider organizations were to get boxed out of the AI revolution, he implies. That goes without saying, but it’s good that someone is kind of saying it.
- Don’t overinvest in large language models or let LLM success cloud your judgment. Instead, explore the world of AI beyond LLMs—and let employees experiment. The pointers are from Robert Blumofe, PhD, chief technology officer with the cloud company Akamai. He offered the tips at a recent MIT event. MIT Sloan School of Management posted coverage of his talk July 21. “In many ways, an LLM is a ridiculously expensive way to solve certain problems,” Blumofe maintains. “It’s rare in an enterprise that you need a trillion-parameter model covering every event in history.” One key lesson to take away from the amazingly cost-effective performance of China’s DeepSeek model, he added, is that “you can do a lot more with less.” Get the rest.
- Headlines too good not to share even though we’re already over ‘optimal’ word limit:
- From AIin.Healthcare’s sibling news outlets:
|
| | | |
| |
| | | | AI can hurt or help the cause of advancing equality of resources, services and outcomes in healthcare. If it’s to do more helping than hurting, the technology must permeate primary care—and do so with robust bias mitigation strategies, comprehensive digital literacy programs and, not least, input from underserved communities. This is the view of researchers in the U.K. who conducted a systematic review of the relevant literature and had their findings published July 18 in the International Journal of Medical Informatics. Primary care physician Ayokunle Osonuga, MBChB, of the University of East Anglia and colleagues found only 15% of healthcare AI tools around the world include community engagement in development. Further, they uncovered, algorithmic bias leads to 17% lower diagnostic accuracy for minority patients. What’s more, close to 30% of patients living in rural areas lack access to AI-equipped healthcare tools. Upon analyzing these statistics and others, the team came up with a number of key observations for moving the needle in the right direction. Here are four. 1. The integration of AI technologies into primary care represents both a tremendous opportunity and a significant responsibility for advancing health equity. “Based on our systematic analysis spanning diverse AI applications, populations and implementation contexts,” Osonuga and co-authors write, “current evidence demonstrates that AI tools—when thoughtfully designed and implemented—can meaningfully address some of the persistent disparities that plague healthcare delivery.” More: ‘From AI-powered risk stratification systems that improve chronic disease management in underserved populations to telemedicine platforms that expand access to care in rural communities, these technologies offer tangible benefits for populations that have historically experienced inferior healthcare.’
2. The potential for AI to exacerbate rather than ameliorate health disparities remains a serious concern that demands sustained attention and proactive mitigation efforts. “Our analysis reveals significant challenges including algorithmic bias affecting up to 24% of diagnostic decisions for racial minorities, digital divide issues excluding 29% of rural populations from AI-enhanced care, and insufficient community engagement in 85% of AI development processes,” the authors report. More: ‘The challenges of algorithmic bias, digital divide issue, and insufficient community engagement are not merely technical problems to be solved through better algorithms or more sophisticated technology. They represent fundamental questions about power, participation, and justice in healthcare that require thoughtful policy responses and genuine commitment to equity principles.’
3. Successful AI implementation for health equity requires a paradigm shift from technology-first to equity-first development approaches. “This shift involves meaningful community engagement from the earliest stages of AI development, robust bias detection and mitigation strategies, comprehensive digital literacy and access programs, and policy frameworks that prioritize equity alongside safety and efficacy,” the researchers write. ‘Without these foundational elements, AI technologies risk becoming another source of healthcare inequality rather than a solution to existing disparities.’
4. The path forward demands collaboration across multiple sectors and stakeholders. These include technology developers, healthcare providers, community organizations, policymakers and the communities most affected by health disparities, Osonuga and colleagues state. More: ‘This collaboration must be genuine and sustained, moving beyond consultation to true partnership in shaping the future of AI-enhanced healthcare delivery.’
The paper is posted in full for free. - In other research news:
- Cleared for market:
- Funding:
|
| | | |
|
| |
|
|
