This past May, ASCO announced its collaboration with Google Cloud to launch the ASCO® Guidelines Assistant, a new interactive tool that allows clinicians to quickly access ASCO’s evidence-based clinical guidelines to facilitate critical clinical decision-making. Developed with Google Cloud’s Vertex Artificial Intelligence (AI) platform and Gemini models, the ASCO Guidelines Assistant provides users with clear citations and sources of all information generated by the tool and enables oncologists to ask follow-up questions about specific treatment recommendations for a particular cancer based on ASCO’s systematic reviews and the work of expert oncologist guideline panels.
The Assistant also allows users to provide feedback that will be used to refine and improve the tool in the future. However, the Assistant is not meant to function as a clinical support tool, according to Clifford A. Hudis, MD, FASCO, FACP, Chief Executive Officer of ASCO and Executive Vice Chair of the Conquer Cancer Foundation. ASCO Guidelines Assistant is available free to ASCO members and can be accessed through asco.org/ga and the ASCO Member app, which can be downloaded on Google Play and the Apple App Store.
In this interview with The ASCO Post, Dr. Hudis describes the goals of the ASCO Guidelines Assistant, how clinicians from both community and academic settings may benefit from the tool’s capabilities, and how this technology is enabling clinicians to deliver better personalized patient care.
Providing Accurate, Reliable Information, Fast
Please talk about how the ASCO Guidelines Assistant can help improve clinical care for patients. What feedback have you gotten so far from ASCO members on the tool’s usefulness?
Let me start with a very simple explanation of the problem we were trying to solve. ASCO has a large number of very high–quality, expert-vetted guidelines that cover a broad range of challenging clinical decisions and stages of care for patients with cancer. However, the way that our guidelines have historically been presented is essentially through journal articles and journal article updates, which are published on an irregular basis when needed and to some degree are constrained by our limited bandwidth.
From a practical point of view, the result of that limitation is that a busy clinician with an acute question in the clinic does not have an efficient and easy way to quickly look up and find an answer to a narrow, specific question about an ASCO guideline recommendation on a particular treatment for a particular cancer. What I mean by that is the clinician would have to recall whether ASCO actually has a guideline on the topic the clinician is interested in retrieving; determine where the answer may be found; and then do a conventional online search if the document isn’t already handy, find it, and then scroll through it to find the answer.
The goal of the ASCO Guidelines Assistant, which includes a chat intermediary, is to quickly surface the answer to a question in a way that establishes reliability and trust.— CLIFFORD A. HUDIS, MD, FASCO, FACP
Tweet this quote
The challenge for us was how to have content from those high-quality, trustworthy, vetted documents be easily discoverable and usable, and that’s where the idea for this program originated. The goal of the ASCO Guidelines Assistant, which includes a chat intermediary, is to quickly surface the answer to a question in a way that establishes reliability and trust. For example, we built this tool to do a couple of things that are a little bit different from the way that most of us interact with various AI platforms, such as ChatGPT, Claude, or Gemini, to name a few.
First, the answers provided by the ASCO Guidelines Assistant are constrained to our guidelines’ documents. So, unlike a broader query in which a large-language AI tool might survey all manner of sources for an answer, we trained our Guidelines Assistant to answer only questions using information contained within our guidelines. And when we don’t have an answer, to say we don’t have an answer, and that’s different from other tools that will almost always give an answer no matter what, regardless of accuracy.
Second, and the subtle and maybe most important aspect of this tool, is that every single one of the answers is referenced. Users can click on the reference link, which I think they should always do; and the source material is then displayed on a panel on the right in a second window, which shows the original reference document and then, highlighted in yellow, the exact sentence or sentences that support the answer provided by the chat.
Taken together, the ASCO Guidelines Assistant is a tool that efficiently exposes the content of our guidelines directly to users, so they can make the most informed clinical decision. And that’s a slightly different process from other large-language AI tools, which give you an answer but not necessarily with supporting, identifiable, or accurate references. Yes, they will all give you an answer, but you really need to see how that answer was sourced to ensure its accuracy.
The goal of the ASCO Guidelines Assistant is to make it easy for clinicians to quickly find an answer but also tell them when we don’t have an answer, so they don’t waste their time.
In terms of feedback from ASCO members, it’s still early days since the Assistant was launched on May 21, 2025. We certainly got feedback soon after the tool was introduced, and some users where quick to point out places where the answers fell short. We asked for feedback on every interaction with the Assistant; using that information, we expect this tool will only get better over time.
Delivering ASCO Guidelines Resources to Every Oncologist
How might this new tool be especially beneficial to community oncologists who have large patient volumes and must have expertise across many cancer types, particularly those practicing in rural communities, where they have limited access to resources and workforce shortages are common?
The utility of the ASCO Guidelines Assistant is broad. You’ve outlined several groups of clinicians that might benefit from its use, but I think the utility of this tool is universal. Basically, the Assistant is beneficial for anyone who wants to know quickly whether ASCO has a guideline or a comment on a specific cancer treatment and where it can be found. So, this tool provides all users with efficiency and clarity, using only trustworthy sources.
Limiting Queries Specific to ASCO’s Guidelines
Are there plans to incorporate other uses for patient care into the ASCO Guidelines Assistant in the future?
Not at the moment, because there are large-language AI tools that use wider bodies of information already available. The key advantage to our Guidelines Assistant is the constraint that it is only able to give answers that are based on ASCO’s vetted, expert-developed guidelines. I think it’s possible that there might be other equally qualified sources admitted to this environment in the future. I wouldn’t rule that out. But at the moment, we are focused on making sure this tool is doing a good job with our guidelines.
Improving Cancer Care Globally
Although this is a time of unprecedented advances in cancer care, the oncology community is facing daunting challenges, including workforce shortages. How might artificial intelligence in the future help fill critical workforce gaps, improve patient outcomes, and reduce physician burnout?
There are a lot of ways to answer your question, and I want to start very broadly. The leaders in technology today on multiple occasions have referred to the introduction of artificial intelligence, which has really exploded in the public awareness over the past 2 years, and the upcoming introduction of artificial general intelligence as arguably the greatest technological revolution in history. I’ve heard AI compared to the creation of the transistor, a fundamental component in modern electronics, and the creation of the Internet. And during the opening session at the 2025 ASCO Annual Meeting, Ruth Porat, President and Chief Investment Officer of Google, suggested that the impact of AI on society will be as transformative as the invention of the steam engine and electricity centuries ago.
I know it’s possible to overhype these important technological advances when we are experiencing them in the moment, but let’s just say the hype is true regarding AI, and so the repercussions on society are beyond broad. I’m optimistic about what this technology offers us as clinicians, now and in the future.
First, I think AI will become useful to a broad range of clinicians practicing in a variety of differently resourced environments to at least be able to access the latest, cutting-edge, clear, and most accurate information on cancer care. They may not always have access to essential resources, such as adequate funding for basic or advanced cancer treatments, but they will at least be able to quickly know what treatment for a specific cancer is optimal. And that knowledge can improve resource allocation and, ultimately, improve cancer care globally.
Second, which gets more to your question about workforce shortages and the challenges oncologists are facing, especially those practicing within the U.S. health-care system, is that I’m optimistic that AI-driven technology can enhance efficiency by taking over time-consuming tasks, such as managing some aspects of patient data recording, discovery, and transmission. Further, I think there is the possibility that this technology can help us begin to turn a corner on efficiency, so we can deliver smarter, faster, and more personalized patient care and, therefore, get back to offering more of the human touch and connection that bring so many of us to oncology in the first place. I imagine a future in which clinicians can face their patients, hold their hands, and spend more time conveying treatment guidance, as well as support and empathy.
Said more plainly, to the degree that AI tools both visible and invisible automate various aspects of clinical care, they should enable clinicians to focus on the parts of care where they can add the most value and impact. I believe there is the possibility that this technology will actually improve the situation for oncologists both here and globally.
DISCLOSURE: Dr. Hudis reported no conflicts of interest.
