Why trusted medicines advice matters more than ever in the AI age
16 Feb 2026
With patients – and even digital tools used inside practices – generating treatment suggestions, having a reliable medicines reference is becoming more important than ever.
In 2024, an Australian study published in the Medical Journal of Australia (MJA) found that 9.9% of adults had used ChatGPT for health information in the previous six months. Of those, 61% had asked at least one higher-risk question that would normally need clinical advice.
In January 2026, OpenAI reported more than 40 million health-related searches were being made on ChatGPT each day.
AI tools are already part of many consultations. Some clinics use digital scribes and ‘copilots’ that help write notes or draft management plans. The benefit is clear – less paperwork and better documentation.
But there are risks. AI can sound confident while being wrong. This can lead to “automation bias”, where people trust an answer simply because it is presented clearly and confidently.
That is why accurate medicines information is critical. When prescribing for patients with multiple conditions, drug interactions, kidney problems, or during pregnancy and breastfeeding, GPs need a dependable reference they can check quickly.
For many doctors, the Australian Medicines Handbook (AMH) is that trusted source. It is regularly updated using Australian expertise and is designed for use at the point of care. In a consult influenced by AI suggestions, AMH can act as the “ground truth” to confirm doses, interactions, warnings and monitoring advice.
Case example: When AI misses the risk
Maria, 76, recently moved to a regional town and saw a new GP. She has heart failure, atrial fibrillation, stage 3 chronic kidney disease, type 2 diabetes, depression and osteoarthritis.
Her medicines list is long. Her My Health Record shows apixaban, sotalol, perindopril, furosemide, metformin and sertraline. She was also given spironolactone during a recent hospital visit, but it was not listed. She regularly takes ibuprofen, fish oil, magnesium and a “natural” sleep product.
Maria presented with symptoms of a urinary tract infection. She showed her GP an AI-generated answer on her phone suggesting trimethoprim, saying it worked for her before.
Before prescribing, the GP paused and checked AMH. The reference highlighted that trimethoprim, when combined with an ACE inhibitor and spironolactone, can significantly increase the risk of high potassium levels – especially in someone with kidney disease.
Regular NSAID use added further risk of kidney injury.
The GP instead chose a safer antibiotic option, ordered urine culture testing and blood tests for kidney function and potassium, and reviewed Maria’s medication list. She was advised to stop NSAIDs, clarify the reason for spironolactone, and bring in her sleep supplement to check for possible interactions with sertraline.
No harm occurred – largely because the risks were predictable and checked against a trusted source.
A simple safety habit
Before prescribing, clinicians are encouraged to quickly check:
-
Dose and patient factors – kidney or liver function, age, weight, frailty
-
Contraindications and cautions – is this safe today?
-
Interactions – prescribed, over-the-counter and complementary medicines
-
Monitoring – what needs to be checked, and when
-
Patient advice – key messages to reduce harm
AI can help draft notes or summarise information. But responsibility for safe prescribing remains with the clinician.
In a time when answers are easy to generate, patient safety depends on consistently checking a trusted medicines reference like AMH and making verification part of everyday practice.
Source: NewsGP