We currently have no oversight and no regulation on AI use in the community, which is the danger because it’s fairly new. Many of us use OpenEvidence for daily use; however, the challenges of using AI to inform clinical practice need regulatory oversight.

The use of OpenEvidence was spread through word of mouth in our community. Someone’s friend from another practice said, “you should use OpenEvidence,” so my colleague used it, and then he told everybody else about it and they started using it, too. Seeing that it has spread through word of mouth and is not promoted on an organization-level is scary to me.

For me, AI is just a tool, but it’s not something I want to depend on. And I think the more I use it, the more there’s a risk of me depending on it or building a dependence that I need to check AI before I check NCCN.

Something that would be helpful for me would be formal guidelines, from ASCO, perhaps, on what to do in this situation. How should I approach OpenEvidence? Is it backed by evidence? Just like how we practice evidence-based medicine and how we choose medications, can this tool be validated, too?

I think we need a little bit more regulation and guidance on how to use it safely for our patients. Having trusted sources like ASCO give their recommendation would be heavily influential.

Also, I think education on AI use in medical practice would be very helpful, not just for myself, but for all community oncologists, given the rapid pace of AI development. It’s hard to keep track of what sources are out there, what sources are credible. What are their pros and cons, and how could that benefit patient care and oncology practice? It’s moving very fast and we don’t have enough time.

Right now, I have an AI scribe. So that’s how I use AI in my practice. Yes, it saves time, but it also doesn’t have the filtering ability of what’s important, what’s not, and capturing the complex discussions we have with our patients in oncology. It’s not like a human scribe who can understand what’s important, what’s not. But it does help with some efficiency here and there.

I think it’d be helpful to know what tools are available and how they’ve been validated. What are their risks and benefits? Is it HIPAA compliant? We use a lot of clinical trials and clinical judgment in oncology to make the best decisions for our patients. Not everybody is a candidate for treatment. Some patients need to be on hospice, and those are all decisions. It’s a complex decision-making skill that we have. How does AI account for all of those factors? I treat a lot of patients with breast cancer. How does AI contribute to mammogram readings? AI is also being used in radiation oncology and pathology for assisting with diagnosis and treatment planning. How does that work? As a tool, for optimizing what we already do, it’s great. But I wouldn’t use it as a tool to help me make decisions, because I think that should still be upon us.

Madiha Naseem, MD, is a medical oncologist and principal investigator at Adventist Health AIS Cancer Center in Bakersfield, California.

Disclaimer: This commentary represents the views of the author and may not necessarily reflect the views of ASCO, Conexiant, or ASCO AI in Oncology.

Do you share similar concerns about AI use in clinical practice? Share your own AI experiences with the oncology community.