Kennedy’s ai push raises new questions for patients 45+

Health and Human Services Secretary Robert F. Kennedy Jr. is promoting artificial intelligence as a way for patients to “fact check” medical advice and gain more control over health decisions. That pitch is landing in a fast-moving debate: some of Kennedy’s allies see AI as a tool to challenge entrenched systems, while many medical experts worry that overselling it could confuse patients and weaken trust in evidence-based care. The Associated Press reported on that split after a recent MAHA summit featuring AI and Silicon Valley companies. (apnews.com)

For adults 45 and older, the issue is practical, not abstract. People in midlife and later life are more likely to manage chronic conditions, see multiple clinicians, take several medications, and navigate portals, lab results, and insurance systems. AI tools are already appearing in those settings, but federal regulators and medical groups are still working through how to evaluate safety, bias, privacy, and accountability. (apnews.com)

What happened

At a MAHA summit, Kennedy praised AI as a way for people to check medical information for themselves, even as some movement figures criticized the appearance of big-tech influence around a cause that began with skepticism of corporations and institutions. The AP also reported that administration officials have increasingly framed AI as a solution for healthcare costs, fraud, workforce shortages, and hospital closures. (apnews.com)

That stance does not mean the government is endorsing unregulated, doctor-free AI care. Instead, the public debate is happening alongside active federal and medical-community efforts to define where AI fits, how it should be monitored, and when human oversight is essential. The FDA, for example, has recently asked for public feedback on regulation of generative AI-enabled medical devices, including risk assessment, premarket evaluation, and postmarket monitoring. (fda.gov)

Why it matters after 45

Adults 45+ are more likely to encounter AI indirectly, even if they never open a chatbot. It may summarize a clinician’s note, help route a message in a patient portal, flag an imaging study, assist with prior authorization, or be built into a device or digital health app. In a best-case setting, those tools could save time and help clinicians focus on harder decisions. But if the system is inaccurate, biased, or poorly supervised, the patient is the one who may face delays, confusion, or misleading information. (apnews.com)

This matters especially for people who may already be managing blood pressure, diabetes, heart disease, cancer surveillance, sleep problems, hearing loss, memory concerns, or medication changes. AI outputs can look polished and authoritative, which makes errors harder to spot than they might be in a rushed appointment or a dense test result. That is one reason medical experts have emphasized that AI should be treated as a tool that supports clinical judgment, not a replacement for it. (pubmed.ncbi.nlm.nih.gov)

What the evidence actually shows

The evidence base is mixed and still evolving. The FDA says AI and machine-learning medical technologies can transform healthcare, but it also notes that performance can change when patient populations, workflows, or inputs differ from the settings used during development. That is why the agency has been researching methods for monitoring AI-enabled devices after they are deployed in real-world care. (fda.gov)

Medical literature also shows that generative AI can produce hallucinations: plausible-sounding but factually wrong output. A recent PubMed-indexed systematic review found 44 studies on methods to reduce hallucinations in healthcare AI and identified strategies such as retrieval systems, human-in-the-loop review, and red-teaming. Another PubMed review found that sociodemographic bias remains a concern in large language models used for medical decision-making. Those findings do not prove that all AI is unsafe, but they do show why supervision and validation matter. (pubmed.ncbi.nlm.nih.gov)

Professional groups have reached similar conclusions. The American College of Physicians has said AI should be used ethically, with attention to privacy, disclosure, fairness, and the patient-physician relationship. A Society of General Internal Medicine statement likewise warned that healthcare uses of generative AI need careful balancing of benefits and unintended consequences, especially in clinical decision-making and the patient-clinician relationship. (pubmed.ncbi.nlm.nih.gov)

What remains uncertain

One unresolved question is how much independent help patients should expect from AI tools. The AP noted that some recent studies suggest AI can outperform doctors in tightly controlled scenarios, but those tests often use detailed written prompts rather than the messy, incomplete information people bring to real life. In real-world use, patients may leave out key details, misunderstand what the tool is saying, or overtrust a confident answer. That gap between lab performance and everyday care is one of the biggest unknowns. (apnews.com)

Another uncertainty is oversight. The FDA’s current discussion paper on generative AI-enabled medical devices is not final policy, and the agency says it is still seeking feedback. That means the boundaries for risk assessment, monitoring, and required evidence remain under active development. In other words, the regulatory system is catching up to technology that is already moving into care settings. (fda.gov)

There is also the broader policy question raised by Kennedy’s message: whether AI will be used mainly to help patients understand established science or to amplify contrarian health claims. Experts quoted by AP argued that useful AI should synthesize the best available evidence, not simply validate whatever a user already believes. That distinction matters because a tool that sounds empowering can still steer people toward poor decisions if its inputs, training data, or design are flawed. (apnews.com)

What to discuss with a clinician

If you are 45 or older and encounter an AI-generated summary, advice chatbot, or automated message about your care, it is reasonable to ask a few simple questions: Is a clinician reviewing this? What source material is the tool using? Has it been tested in people like me? And what should I do if the output seems wrong or incomplete? Those questions are especially important if the result affects medication changes, a new diagnosis, a screening test, or follow-up after abnormal labs or imaging. (fda.gov)

If something feels urgent, do not wait for an AI response. Seek prompt medical care for symptoms such as chest pain, trouble breathing, stroke-like symptoms, severe allergic reactions, confusion, or any rapidly worsening condition. AI can be a support tool, but it is not a substitute for timely clinical judgment or emergency evaluation. (pubmed.ncbi.nlm.nih.gov)

For now, the most realistic view is the least dramatic one: AI may help healthcare work faster and make some information easier to navigate, but its reliability depends on the setting, the oversight, and the evidence behind it. For adults 45+, the best takeaway is not to fear AI or trust it blindly, but to treat it as one more tool that should be checked against sound medical practice. (apnews.com)