Introducing ASCO AI in Oncology: A Conversation With Clifford A. Hudis, MD, FACP, FASCO

In February, ASCO and Conexiant launched ASCO AI in Oncology, a digital platform dedicated to understanding how AI is impacting cancer care.

“Our goal with this hub is to empower oncology professionals with knowledge and the tools to adapt to a rapidly changing clinical and research environment to improve patient care and to ensure that ASCO leads in this transformative technology,” said Clifford A. Hudis, MD, FACP, FASCO, ASCO’s Chief Executive Officer and Executive Vice Chair of Conquer Cancer.

Empowering Clinicians to Adapt to a Rapidly Changing AI Environment

The ASCO AI in Oncology hub is user-friendly with separate key topic categories to help clinicians find the information they need quickly in such areas as Diagnostics & Imaging, Decision-Making Support, Prognostic and Predictive Models, Drug Discovery and Clinical Trials, and Cancer Types, among others. The platform also offers users insights from oncologists and researchers from around the world about their experiences integrating AI into their clinical care and research; timely articles about regulatory concerns regarding data privacy, patient consent, algorithmic transparency, bias mitigation, efficacy and equity; links to the latest research in AI use in oncology; and reporting on important meetings covering AI in oncology.

The platform will also directly link to the AI-powered ASCO Guidelines Assistant and ASCO’s Guiding Principles to Responsible AI Use, as well as a glossary of AI terms. Additional functionality and topics, including patient-related content will be added in the coming months.

In this wide-ranging interview, Dr. Hudis discussed how ASCO is supporting members in their adoption of AI technology to advance patient care, how the Society is incorporating AI literacy modules into its meetings, and the regulations imposed so far to guard against deceptive practices in AI use in health-care settings.

Supporting Oncologists to Advance Patient Care

Please talk about why ASCO and Conexiant are launching the ASCO AI in Oncology platform; and how it will benefit ASCO members in clinical care and in research?

The reason we are launching this platform is that there is an increasing demand for AI in oncology. Artificial intelligence is probably the most rapidly developed and deployed technology that any of us have seen in our lifetime. Its early impact on clinical care is profound and we want this hub to help our community understand what’s coming our way and to begin adapting and adopting this technology to take full advantage of its possibilities to support oncologists and advance patient care.

The reason we are launching this platform is that there is an increasing demand for AI in oncology. Artificial intelligence is probably the most rapidly developed and deployed technology that any of us have seen in our lifetime. Its early impact on clinical care is profound and we want this hub to help our community understand what’s coming our way and to begin adapting and adopting this technology to take full advantage of its possibilities to support oncologists and advance patient care.
- Clifford A. Hudis, MD, FACP, FASCO

Using AI to Help Match Patients to Clinical Trials

What has been the real-world impact so far of AI use in oncology care; and how is AI being adapted into clinical care and research?

This is a much more nuanced and complicated question than it might seem at first, because AI in a rudimentary way has certainly been part of clinical care for many years. What’s happening now, at least broadly, is taking place on three fronts. The first is deployment of AI that is largely or entirely invisible to clinicians and patients in terms of all the back-office operations, including in scheduling and resource apportioning.

The second big area is in overtly providing support to clinicians, and even in that regard, we’re in the earliest days of this technology helping clinicians identify opportunities for better patient care, and also more efficiently finding needed advice in treatment guidelines. Most of these tools are already upon us, but these are still early days. What’s coming in the short-term future is going to be even more supportive than what we have now.

And the third front, which is also not yet available, but soon will be, and this will be largely hidden to the patient, is the part of the research enterprise that is going to benefit from the rapid acceleration of the evolution of AI. This will be in terms of regulatory filings from drug discovery or the design of new molecules for matching and enrolling patients to appropriate clinical trials and enhancing clinical trial design.

These are the three places where we are beginning to see meaningful clinical and research impact: it’s office operations in a very businesslike way, it’s clinical care decision-making and delivery, and its research innovation.

Integrating AI Into Every Aspect of Clinical Care

What feedback are you getting from ASCO members on how they are using AI in their practices, and are there concerns among members about this technology; how is AI enhancing or complicating clinical cancer care?

The AI revolution is so pervasive in terms of its impact on society in general, in some ways its impact on medicine or in health care has been muted. AI is becoming part of every aspect of our life.

I’ll give you a couple of examples: I don’t think there are too many people at this point who aren’t at least dabbling in one of the AI tools being used in everyday life, whether it is with ChatGPT, Gemini, or Anthropic. Number two, the uptake for physicians in particular, of some of the AI tools—I’m thinking here specifically of OpenEvidence, an AI-powered medical search platform that provides evidence-based clinical decision support for physicians—has been more rapid that any other uptake that I think has ever been documented.

I may be wrong, but I think the simple majority of the roughly 1.1 million licensed clinicians in the United States, and probably more at this point, have at least explored OpenEvidence to ask a medical question or to write a prior authorization request, for example.

This is all to say that, yes, we hear comments and we hear lots of thinking about this technology’s increasingly sophisticated applications in medicine and oncology. And certainly, some members have casually expressed concern about what these advancements will mean both in terms of the onboarding acquisition skills needed to understand the basic principles of AI to interpret AI-generated data, as well as the maintenance skills to critically judge machine outputs.

But on the other hand, this is a little bit like talking about the launch of email and the internet decades ago. AI in oncology is a force that will not be stopped, and I think there is acceptance of that fact.

Preparing Clinicians for the AI Revolution

Given the fast adoption of AI in oncology, what has to happen in medical education to prepare clinicians for the use of large language AI models?

I have a slightly different view from some others about this issue because the integration of AI into oncology clinical care and research is going to be a ubiquitous reality in the near future, and clinicians and researchers may not be thinking about or aware of the fact that they are relying on a large language model for some of their work. In 2025/2026, the European Society for Medical Oncology (ESMO) and the United States Food and Drug Administration, established a basic framework for AI in drug development, and I think these guides are both really important. The ESMO framework, called the Basic Requirements for AI-based Biomarkers (EBAI), offers actionable guidance for developers, clinicians, and regulators to support safe and effective AI integration.

But I honestly believe that all of this technology is largely going to be below the surface when it comes to the day-to-day lives of most clinicians, and perhaps not even visible. Everybody needs to be aware of these tools the same way we need to be aware of the idea of electricity or the internal combustible engine, even if we don’t fully understand how they work anymore.

Understanding the Value of AI-Driven Clinical Care and Research

How is ASCO incorporating AI literacy modules tailored to the needs of oncologists into its conferences and in the sessions at the Annual Meeting?

In a couple of ways. On the scientific side, there are abstracts that are submitted to the many thematic meetings and the Annual Meeting that demonstrate the utility or value or impact of the various approaches to AI-driven clinical care and research. And then, from a practical point of view, there will be discussions focused on the uses of AI in oncology, and the responsibility that clinicians have to be critical and thoughtful in picking those outputs.

To be very clear, I’ve used this example before, but I wanted to think about automation in the context of medicine. It’s always easy for me to go to automobile analogies to express my point. Right now, if you buy a car, even a mid-priced car, most will have some degree of an automatic braking system, and most will have lane detection sensors to alert drivers when they are drifting out of their lane. But even with this technology, drivers still have the perception of autonomy. They’re driving the car, braking, and accelerating. However, when they find themselves in a situation that is possibly unsafe, the car, to a forceful degree, nudges them back to keep them in the right lane or to autobrake if a basketball suddenly comes bouncing into their lane to avoid hitting a child.

Now, let’s think about AI in the context of the electronic medical record and clinical care. The future we’re barreling toward is the embedding of all of these tools into electronic medical records, so that autonomous, independent clinicians are, in fact, thinking, testing themselves, and making decisions. But, they are increasingly protected by the guardrails that are in the background of this technology making sure that what they are recommending for their patients fits within standard guidelines. That doesn’t mean that a smart and experienced clinician won’t rightly deviate from those guardrails some small percentage of the time, but when they do, they are going to have to be purposeful about it and able to clearly defend their reasoning. This strikes me as the likeliest place where we end up.

Ensuring Equitable Care for All Patients With Cancer

Although AI is showing enormous potential in improving outcomes for patients with cancer, clinical trial data show that it’s unclear how these tools will generalize to a more diverse population to ensure equitable care for all patients with cancer. How can potential biases in this technology be addressed and eliminated?

Let’s start with a different take on that question, because there are well recognized disparities in access to oncology care and the impact that has on outcomes. One result from the ubiquitous use of AI tools is that inequity in cancer care will become even more glaring when a patient is not able to obtain the same resources and outcome as other patients, because AI tools are going to define a standard of care and make it harder to deviate from it.

The rising tide lifting all boats is what I’m getting at here. I think the question you’re asking is a slightly different one, and that is, how do we protect against these tools being wrong or misleading because of input bias in AI? And I would answer that question in two ways.

Number one is, this brings us back to the fact that, at least for now, the clinician is still completely responsible for patient treatment decision-making and care. Clinicians are using these AI tools as a way to test themselves, of checking to see if they’ve missed something in a patient’s care. But clinicians are ultimately responsible for their patient’s care, and that means that they have to be critical readers of the answers they get from AI, rather than simply accepting those answers as fact.

The flip side of that equation is that with greater liquidity, connection, and accessibility of the data, I think in the long run, we can help solve the problem of inequity in care by not excluding any patient from the datasets.

Raising the Quality of Care

How do you expect AI technology to improve patient care and what are your concerns?

I expect this technology to improve patient care by making it easier for clinicians, no matter where they are, to have access to the latest and most accurate and contextually appropriate information in the patient setting. So, again, I see this technology as a rising tide lifting all boats.

Once more, I’ll use the car analogy to make my point. When I think about Waymo cars in California, for example, everyone is so concerned about the risks and dangers of these autonomous ride-hailing cars, but the truth is that expecting perfection from them is unrealistic. We don’t expect perfection from our hired human drivers, or from ourselves, either.

So, the real question becomes, how do these tools compare to the current standards in cancer care? And, I think in that regard, they are going to raise the quality of care, because humans are capable of making mistakes. But the combination of a smart human and a smart AI tool should drive us to greater heights of quality.

This gives me a chance to emphasize what I said earlier, which is, at least for the foreseeable future, clinicians using these tools are responsible for the output, not the tools.

Now, as of January 2026, Utah became the first state to launch a pilot program allowing an autonomous AI health platform to independently renew certain prescription medications without a human doctor’s review. And that violates what I just said because once that happens, clinicians are no longer in the loop, at least not in a traditional manner.

So, we are going to continue to see examples of this kind of progress with AI, and we will have to constantly adapt to this progress. But this is just a replacement for other systems that provide services to the health-care system as a whole.

Instituting Guardrails to Protect Patients

So, will we need some kind of regulation to ensure that a machine isn’t doling out cancer medications without an oncologist knowing about it?

I think that’s right, and we’re in the early days of instituting more guardrails around this technology. Last year, California passed a law banning AI from using terms implying health-care licensure or certification in a health-care profession and protecting consumers from misleading representations in AI-generated health advice, care reports, or assessments. So, we are starting to see more regulations that make it clear clinicians can’t solely rely on AI for decision-making, and hold them accountable when they do. Clearly, more clarity is needed as we move forward.

DISCLOSURE: Dr. Hudis has no conflicts of interest to declare.

ASCO AI in Oncology is published by Conexiant under a license arrangement with the American Society of Clinical Oncology, Inc. (ASCO^®). The ideas and opinions expressed in ASCO AI in Oncology do not necessarily reflect those of Conexiant or ASCO. For more information, see Policies.