AI & HEALTH · FOR PATIENTS

AI Symptom Checkers Compared

'Symptom checker' now means two different things: dedicated apps and websites built to ask about symptoms step by step, and general chatbots that answer symptom questions in open conversation. They feel similar but work differently, fail differently, and deserve different levels of trust. This page compares how each reaches its suggestions — without league tables, because no honest ranking exists without standardised testing, which is why we are building our own.

By Dr. Mohamed Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine

✓ Medically reviewed by Mohamed Magdy Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine · Board certified in Internal Medicine (ABIM) · Last reviewed: November 2026

AI & Health is the MedicalVideos.com section on artificial intelligence in health care. We recommend no commercial product; where we test AI systems, we publish our method and our full results.

In short: Dedicated symptom checkers follow a fixed question path built for triage — how urgently you should seek care — and are usually cautious by design. General chatbots are flexible explainers but improvise: no fixed safety path, no guaranteed triage, and answers that shift with how you phrase the question. Neither diagnoses you. Use checkers for urgency guidance, chatbots for explanation, and clinicians for decisions.

The guide

The picture today

A dedicated symptom checker is built around a structured interview. It asks a defined sequence of questions, follows branching logic written or approved by clinicians, and ends with two outputs: a list of possible causes and, most importantly, a triage recommendation — self-care, see a clinician soon, or seek urgent care. Because the pathway is fixed, the same answers produce the same output, and the tool can be tested end to end before release. Many are regulated as medical devices in some jurisdictions, which imposes design and monitoring requirements a general chatbot does not face.

Strengths and limits

A general chatbot has no fixed pathway. It generates its reply from your wording, so the questions it asks, the causes it lists, and the urgency it conveys all depend on phrasing, order, and tone — ask twice, differently, and you may get different advice. That flexibility is its strength for explanation and its weakness for triage: nothing forces it to ask the one question that separates a harmless cause from a dangerous one. Independent evaluations of both kinds of tool have found wide variation, which is why this page makes no accuracy claims for any product.

In practice

Choose the tool by the job. If the question is 'how urgently should I be seen?', a dedicated checker built for triage is the more appropriate design, and its caution is a feature, not a flaw — follow the more urgent advice when tools disagree. If the question is 'help me understand this condition or prepare for my appointment', a chatbot's explanations are more useful. With either tool, answer fully and honestly, do not steer it toward the answer you want, and remember that a list of 'possible causes' is ordered by the tool's logic, not by what is actually wrong with you.

Red flags

Whichever tool you use, its output never outranks your own judgement or a clinician's. If a checker or chatbot suggests self-care but you feel seriously unwell, are getting worse, or have any emergency symptom — chest pain, breathing difficulty, stroke signs, severe bleeding, confusion — seek emergency care anyway. Tools also underperform for young children, pregnancy, frail older adults, and people with weakened immune systems; for those groups, lower your threshold and speak to a clinician directly.

Common questions

Which symptom checker is the most accurate?

There is no trustworthy public league table. Studies use different symptom sets and scoring, and often test versions that have since changed, so results do not transfer to the tool in front of you today. We are building a standardised, published test of our own; until results exist, we will not rank products.

Why do two tools give me different answers?

Because they are built differently. A dedicated checker follows its fixed pathway and weights urgency conservatively. A chatbot improvises from your wording. Different question sets, different logic, and different caution settings produce different outputs — one more reason no tool's list of causes should be read as your diagnosis.

Are symptom checkers regulated?

Some are, depending on what they claim and where they are sold. Tools offering diagnosis or triage may fall under medical-device rules in the United States, Europe, or elsewhere, while general chatbots often sit outside those rules. Regulation is a floor, not a guarantee of accuracy.

Can a symptom checker miss something serious?

Yes. Any tool working from typed answers can miss what an examination would catch. That is why the standing rule is: worsening symptoms, emergency symptoms, or your own strong sense that something is wrong always override a reassuring screen.

More in For Patients

Related

Sources

This page is educational content about artificial intelligence and health. It is not medical advice and does not replace your own clinician's judgment. Never use an AI tool in an emergency: if you think you may be having a medical emergency, call your local emergency number (911 in the United States) or go to the nearest emergency department. No tool output on this page or anywhere else replaces triage by a qualified clinician.