Questions Patients Ask About AI (and How Your Clinic Should Answer Them)
When a clinic introduces AI, whether it is a voice agent answering the phones or software supporting documentation, patients have questions. Most never ask them out loud. They just form quiet conclusions based on whatever they can guess.
That is the real risk of AI in a health setting. Not the questions. The unanswered ones.
Below are the questions patients actually have, based on what comes up when Ontario clinics talk to their communities about AI, along with answers a clinic can adapt and use. If you are a clinic owner, consider this a starting draft for your own patient FAQ. If you are a patient, these are the questions you have every right to ask, and the answers a well-run clinic should be able to give you.
"Am I talking to a robot?"
If your clinic uses AI properly, you will never have to wonder. The AI will tell you.
Our standard, and the standard we believe every clinic should hold, is that any AI involved in a phone call or message identifies itself at the start. No guessing games, no gotcha moment later. If you ask a clinic whether AI was involved in a communication you received, you deserve a straight answer.
"Can I still talk to a real person?"
Yes, and this is the single most important commitment a clinic can make. During clinic hours, asking for a person should get you a person, without friction. The AI never stands between you and the care team.
After hours and on weekends, the honest answer is that the clinic’s humans have gone home, the same as they did before AI existed. What changes is what happens to your call. Instead of a full voicemail box, the AI can answer, take an accurate message, tell you exactly when the team will get back to you, and flag anything urgent so it is seen first thing. And for anything that cannot wait, it should immediately point you to the clinic’s after-hours protocol, whether that is an on-call service, an urgent care clinic, or 911.
The commitment is not that a human is available every hour of every day. No clinic has ever promised that. The commitment is that the AI is never a wall: you always know how to reach a person and when you will hear from one. If a clinic’s AI has no path to a human at all, that is not responsible automation, and you are right to object to it.
"Is the AI making decisions about my health?"
No, and it never should. AI in a responsible clinic supports administrative work: scheduling, reminders, messages, documentation support. It does not diagnose, does not make treatment decisions, and does not replace a conversation with your doctor or care team. Any clinical question you raise with an AI assistant should be routed to a human, every time.
"What happens to what I say?"
In Ontario, your personal health information is protected under the Personal Health Information Protection Act (PHIPA), and that protection follows your information into any AI tool the clinic uses. A note, message, or summary produced with AI support is protected the same way a paper record is.
Clinics are also responsible for their technology vendors. Any company handling health information on a clinic’s behalf must operate under a written agreement and carry the same obligations the clinic does. You can ask your clinic whether such an agreement exists. A prepared clinic will say yes without hesitation.
Two more questions worth asking: where does the data live, and how long does it stay there? Many of the technology platforms behind AI tools are not Canadian companies, which means call data can be processed or stored outside Canada. That is permitted under PHIPA with the right agreements and safeguards in place, but a clinic should be able to tell you it knows where its data goes. Retention is the other half. Our own practice is data minimization: anything that belongs in your health record goes to the clinic’s own systems, where it is kept according to Ontario’s record retention requirements, and data sitting in the technology pipeline, such as call recordings and transcripts, is deleted on a schedule of between three months and one year, depending on the type of data. The less data sitting in third-party systems, and the shorter it sits there, the smaller the risk.
"Is my information being used to train the AI?"
This is one of the sharpest questions a patient can ask, and clinics should welcome it. The answer at a well-governed clinic should be no: vendor agreements should prohibit using patient information to train the vendor’s models. If a clinic cannot answer this question, it has homework to do. We cover what clinics should demand from vendors in our plain-language guide to how PHIPA applies to AI.
"Why is the clinic doing this at all?"
Fair question, and the honest answer is capacity. Clinic phone lines miss calls every day, not because staff do not care but because they are caring for the patients in front of them. After hours, nobody answers at all. AI handles the routine calls, the bookings, the reminders, and the messages so that the humans have more time for the work only humans can do. Done right, you should experience it as a clinic that is easier to reach, not one that is harder to talk to.
"What if I just do not want to deal with AI?"
Then you should not have to. Ask for a person, and a responsible clinic will connect you with one, without friction and without making you feel difficult for asking. Your comfort is not an obstacle to the clinic’s technology plans. It is the measure of whether those plans are working.
"Where can I learn more or raise a concern?"
Start with your clinic. Ask them directly, and expect plain answers rather than brochure language. Beyond the clinic, every Ontarian has rights under PHIPA, including the right to access your health information and to raise privacy concerns with the Information and Privacy Commissioner of Ontario.
One Ontario clinic recently set a strong example by publishing a plain-language page answering these questions before patients had to ask, an approach we broke down in our article on telling patients about AI first. More clinics should follow it.
For clinic owners reading this
Notice something about every answer above: none of them are hard. They are only hard if you have not done the underlying work, the privacy review, the vendor agreement, the disclosure policy, the human handoff path.
Do the work, and this FAQ writes itself. Skip the work, and every one of these questions becomes a moment your patients catch you unprepared. A patient-facing FAQ like this one belongs in every clinic’s AI adoption toolkit, next to the transparency page and the staff training. If you want help building yours, that is exactly what we do.
We automate the work, not the relationship.
Ajax Web AI helps clinics and local businesses across Durham Region and the GTA adopt AI voice agents and automation responsibly, including patient-facing FAQs and transparency pages as standard deliverables. Contact us to build yours.