My View – Advances in Health AI

VIEW - Chris Paton, associate professor in Health AI, Auckland University 

At the Artificial Intelligence in Medicine and Imaging (AIMI) conference held at Stanford University in June 2025, the health lead for OpenAI told the audience that their AI models are advancing faster than ever. Healthcare is a priority for them.  

The much-discussed plateauing of LLM performance is over and new approaches are producing much better results. Multi-modal AI is mind-blowingly good. 

Clinical questions

Clinicians in the audience asked pertinent questions: Am I liable when the AI makes a mistake and my patient sues me? Will patients prefer to see an AI rather than a real doctor? We do not yet know - there is little regulation and this is new territory. 

Patients will vote with their feet: it will be difficult to beat the convenience of chatting to a personalised AI avatar over waiting weeks to see a human doctor. The AI will appear to have amazing medical knowledge and, at some point, may be able to reason as well or better than a human doctor.

AI research

I was invited to be a panellist discussing publishing AI and digital health research. 

I have recently started as editor-in-chief of a new journal for the BMJ group called BMJ Digital Health & AI. Our panel discussed the need for journals to keep up with AI research, to publish important studies on AI competence as quickly as possible, and how digital health research should follow the principals of evidence-based medicine and open science. 

Government regulation and professional standards on the use of AI technology should be guided by high quality research which we can contribute to by publishing papers and providing a forum for scholarly discussion.

AI is coming for healthcare

Other panels discussed venture capital funding for health AI which seems to be growing in line with the technological developments, and how academic medical centres such as Stanford Medicine have developed policies and monitoring processes to ensure AI use is as safe and effective as possible.

The biggest driver of recent AI adoption is the efficiency of using AI scribing for documenting consultations. However, the lightening talks session demonstrated that AI is transforming medical imaging and becoming increasingly embedded across many other aspects of healthcare.

As usual, the really interesting conversations took place at the conference dinner in downtown Palo Alto. Technologists, clinicians, VCs and academics discussed their projects and latest ideas and tried to come to terms with the fact that AI is coming for healthcare. 

The technology is developing fast. We now need to figure out how to use it effectively while keeping patients and their data safe.

Chris Paton will be giving a public lecture about AI in healthcare at the Liggins Institute on the 14th August at 6pm. Register here: https://www.eventbrite.co.nz/e/one-size-fits-all-medicine-is-killing-us-how-precision-health-can-save-us-tickets-1458875258169

Read more VIEWS

Return to eHealthNews.nz home page