Why Do More People Feel Ill Today?

A 5-minute read on why knowing more about your body can leave you feeling less sure

Psychology Technology Health & well-being

1The Paradox of the Informed Human

A person wakes up and checks their phone. They read that sugar is dangerous, that a common food might cause inflammation, that a vitamin deficiency may be more widespread than they thought. Later, they ask an AI assistant about a faint tingling in their arm. The answer lists several explanations, including a few serious ones. They feel they’ve learned something important. But instead of feeling reassured, they begin monitoring their body, hunting for more signs, and asking more questions.

This isn’t everyone’s experience. But in an information-saturated society, it’s an increasingly plausible one.

Are we becoming more knowledgeable about our health, or just more conscious of everything that could go wrong?

2From the Information Age to the Overinformed Human

The internet made medical information easy to find. Conversational AI made it easy to personalize. Where people once searched a single symptom and skimmed a few articles, they can now describe several symptoms, add their age and habits, and receive an elaborate answer with possible diseases, risk factors, and suggested tests. Three things get blurred along the way:

InformationWhat could be true in general.

InterpretationWhat it means in a particular context.

DiagnosisA clinical conclusion reached through proper assessment of the individual.

They are not interchangeable. A detailed answer doesn’t prove that an AI system has identified your condition, and more possibilities can create the feeling of understanding without improving certainty.

Evidence Reviews of information overload, largely from business and accounting settings, suggest decision accuracy tends to fall once information exceeds what people can process.⁸ Interpretation Applying that to health decisions is an extrapolation.

3When Hypothetical Possibilities Begin to Feel Personal

AI answers can feel unusually personal because they use the details you give and respond in conversational language. Ask about fatigue and you may read about poor sleep, nutritional deficiencies, stress, thyroid disorders, and diabetes. Some may fit you. Others may be unlikely or irrelevant.

There’s a psychological gap between reading that a condition can cause a symptom and believing it might explain yours. A personalized-sounding answer is not a personalized diagnosis: relevance depends on medical history, examination, test results, and clinical judgment.

An AI assistant can help you explore what might explain a symptom, but a list of possibilities is not a list of conditions you have.

That doesn’t mean every AI response is inaccurate. The issue is how uncertain information is interpreted and acted on.

Evidence In a 2026 randomized study of 1,298 UK participants, language models tested alone identified relevant conditions in 94.9% of scenarios. But people using the same models identified relevant conditions in fewer than 34.5% of cases, no better than a control group.⁶ Participants often didn’t know what information the model needed, and responses mixed good and poor suggestions.⁷ Older symptom checkers listed the correct diagnosis first in 34% of cases.⁵

4The Anxiety Loop: How More Answers Create More Questions

You notice a physical sensation
You search or ask an AI assistant
You meet several possible explanations
You watch your body more closely
New sensations prompt new searches
Brief relief, lingering worry

…and the cycle begins again

Health anxiety is a recognized phenomenon, and online searching can feed it. Researchers use the term cyberchondria for excessive online health searching linked to heightened distress. It’s a descriptive label, not a formal diagnosis, and it’s associated with health anxiety and intolerance of uncertainty.¹² Checking brings quick relief, but the relief fades and the habit strengthens.³ Focusing on one body area also makes us notice sensations we’d normally ignore.⁴

None of this means symptoms are imaginary. Genuine conditions and anxiety can coexist, and persistent symptoms deserve proper assessment.

Interpretation Whether conversational AI creates a distinct version of this loop hasn’t been established. It’s a hypothesis worth testing, not a new disorder.

5Why the Favorable Answer Beats the Reliable One

People want certainty, convenience, privacy, and reassurance. A reliable assessment can mean time, money, appointments, tests, and follow-up. An AI assistant is instant, and may offer an explanation that feels manageable. Another response might list several serious possibilities and leave you frightened.

Feels favorable

Instant and private
Reassuring or alarming
Judged by emotion

Supported by evidence

Verified against sources
Placed in your context
Judged by reliability

The most comforting answer isn’t necessarily correct, and the most alarming isn’t necessarily more accurate. In fact, studies of symptom checkers found their triage advice tended to be risk-averse, urging care even when self-care was reasonable.⁵

The real problem isn’t that professional care is always inaccessible, or that AI is always unreliable. It’s that convenience and emotional satisfaction can become substitutes for verification.

6Food, Lifestyle, and Constant Self-Surveillance

The same dynamic extends to the wider culture of health optimization. People may become preoccupied with calories, ingredients, supplements, inflammation, hormones, sleep scores, and exercise metrics, and with the possibility that ordinary foods or habits are quietly damaging them.

Contradictory advice and endless risk information can turn healthy behaviors into rigid rules and everyday choices into sources of guilt. Nutrition awareness, preventive care, and evidence-based lifestyle change are valuable. The problem is feeling compelled to investigate and control every possible risk.

Overinformed but under-contextualized

7How to Be Informed Without Being Consumed

Use AI to prepare questions for a clinician, understand terminology, and explore general concepts.

Treat AI-generated possibilities as starting points for inquiry, not confirmed explanations.

Don’t re-ask the same question in different ways just to find certainty.

Rely on trusted medical sources, and seek assessment when symptoms persist, worsen, or are urgent.

Accept that some uncertainty is unavoidable and not every sensation signals disease.

Set limits on health searches when they raise distress without adding useful information.

Important: none of this means ignoring symptoms. Appropriate monitoring, prescribed treatment, and necessary tests still matter, and sudden or severe symptoms need urgent care.

8Knowledge Should Give Us Perspective, Not Permanent Fear

Information exists to help us understand the world and make better decisions. When every possible risk feels personally threatening, it loses that empowering function.

The modern challenge isn’t simply acquiring more knowledge. It’s learning to tell possibilities from probabilities, general information from individual assessment, and useful awareness from anxiety-driven checking.

Perhaps the most important skill in the age of AI is not asking more questions, but learning which answers deserve our trust, and when a possibility is not yet a reason to worry.

References

Vismara M, et al. Is cyberchondria a new transdiagnostic digital compulsive syndrome? A systematic review of the evidence. Compr Psychiatry. 2020;99. doi:10.1016/j.comppsych.2020.152167
Starcevic V, Berle D, Arnáez S. Recent insights into cyberchondria. Curr Psychiatry Rep. 2020;22. doi:10.1007/s11920-020-01179-8
Osborne DWS, Williams CJ. Excessive reassurance-seeking. Adv Psychiatr Treat. doi:10.1192/apt.bp.111.009761
Cumbria, Northumberland, Tyne and Wear NHS Foundation Trust. Health anxiety self-help guide.
Semigran HL, et al. Evaluation of symptom checkers for self diagnosis and triage: audit study. BMJ. 2015;351:h3480. doi:10.1136/bmj.h3480
Bean AM, et al. Reliability of LLMs as medical assistants for the general public: a randomized preregistered study. Nat Med. 2026;32(2):609-615. doi:10.1038/s41591-025-04074-y
University of Oxford. New study warns of risks in AI chatbots giving medical advice. 10 February 2026.
Roetzel PG. Information overload in the information age. Business Research. 2019;12(2):479-522. doi:10.1007/s40685-018-0069-z; Hartmann M, Weißenberger BE. Information overload research in accounting. Manag Rev Q. 2024;74(3):1619-1667. doi:10.1007/s11301-023-00343-7