by Elaine Hamm, PhD, President and CEO, Ascend BioVentures
I had the honor to help host the Alliance for Artificial Intelligence in Healthcare’s Second Annual Members Meeting, hosted at the campus of Genentech in San Francisco. It was an intense few days and though not the official summary from the AAIH, I wanted to share my “musings” fueled by the strong coffee in the AA airport lounge.
Brought to you by the letter T:
We kicked off the conference with the CEO of Genentech, Ashley Magargee and Genentech’s approach to AI in pharma. The Genentech approach to AI is focused on the letter “T” the shape of which indicates Breadth and Depth and the use of AI, not in just one area of their company, but multiple. From discovery to manufacturing to clinical, the approach to using technology to leverage quality data and a deep understanding of biology across the value chain is one the Genentech has fully embraced (and has done so for a long time…more on that later). In a fireside chat, Dr. Magargee emphasized the need to focus on explainability and simplicity to help build trust and to across the value chain of healthcare. Key to this is the human side and the need to keep humans integrated into the lab in the loop.
Inspirational but grounding
We had several speakers who placed before us a vision of a healthcare future. Dr. Daniel Kraft’s explosive lightning talk (I think maybe 200 slides in 20 min?) presented the possibilities of prevention not treatment, unified databases, and even personalized formularies (ie…take one pill designed for you for all of your ailments) talk. Digital twins and AI hospitals were discussed. But the low hanging fruit and natural use of AI is first where the boring, repetitive tasks are. Billing, Patient access, contracts, and areas where lack of staffing is felt the most. We have quite a bit to go on some of the hypothetical concepts and a number of challenges before us (a looming question was what is Healthcare’s “AI Strawberry issue?” IYKYK). But have no fear, people won’t be replaced by AI any time soon. But grumpy people who scream at us to “get off their yard” and refuse to think about AI, might be. Sadly, learning is easy. Unlearning is harder.
Controversies? Er…Opportunities?
Our Keynote Speaker, Dr. Matthew Lungren, Chief Data Science Officer of Microsoft Health reminded us that if you aren’t losing sleep, you haven’t come across really good AI. (Side note… This was made all too real when my own voice-but not my voice but kinda my voice-was played over the speakers as part of an ai generated portion of his talk). He also highlighted the evidence of confidentially wrong” medical advice was a concern with medical AI (but that we have seen that we some human doctors as well). He went on to discuss how AI bots were also perceived to be MORE empathetic than their human counterparts and that maybe the bots could teach us a few things on patient interaction. As with any conversation around AI, controversial topics were discussed. Ethics, first to market, standards, innovation, discovery, regulation are all part of the tennis match of AI. Perhaps the liveliest panel was that of the VC panel when Dr. Vas Bailey casually questioned whether countries with less regulation have more innovation and first mover advantage (bringing up China’s AI hospital). Yet this panel was hot on the heels after an entire hour around the need for standards across the industry during a breakout session focused on creating an AI playbook as well as the quantity vs quality data debate. A tale as old as time, the balance of innovation and ethics continues.
Always ripe with controversy is the area of AI and Medical claims. It is quite intriguing to think of the ambient note taker in the exam room being able to help doctors and hospitals not just record accurate patient records without having them having to transcribe and upload BUT also create documents that would help with accurate and (and more reimbursable) medical billing claims. After all, insurance companies are already using AI as a way to help them say no to claims (with the class action lawsuits to go with it). And now, patients also can use AI to help with their denials as well (check out Holden Karau’s Fight Health Insurance platform). But rather than a healthcare billing Thunderdome betwixt hospitals, insurance, and patients…could we not use AI to consider negotiating terms to help ensure that everyone benefits and patients are made better and not sicker and poorer?
Better late than never. A common statement was that if you started thinking about AI three years ago you are behind. If you haven’t started yet, you were REALLY behind. BUT implementation of data management and an emphasis on data quality in your organization could help you leapfrog when the time is right for your AI strategy. (Don’t fret….the AAIH is creating an AI playbook to help guide the process of how and why to invest based on your desired goals….stay tuned on that). In fact, one of the main reasons I joined the AAIH was BECAUSE I didn’t understand AI and I felt behind (nothing like drinking from the firehose to orient you)!
And now for the most important takeaway. What may surprise you at an AI conference was that humans were the most talked about at this conference. Never was it clearer that where AI shines the most is as a helper and tool to humans and not as our replacement.
We need our own Virgil to help us guide us through the diverse, overwhelming, and vast landscape that is healthcare and AI. This was clear from our healthcare panelists Ryan Hildebrand of LCMC Health and Dr. Jud Schneider of NASHBIO as they discussed the need for organizations to designate (human) guides to help navigate large, unstructured, and complicated data sets. Also discussed was the value of creating “Tool Teams” at non-AI companies to help weed out need vs nice to have and sort out value over hype. At Tulane Med we are working hard to do this as we are inundated with requests to try AI tools for FREE (so long as they get our valuable data).
But poignantly, the potential to use of AI to help alleviate the pain, frustration, and burnout that is all too real for our healthcare personnel. As revealed by Dr. Lungren, medical staff and doctors can spend up to 24 hours on paperwork and admin. The greatest status symbol may go from having the best car in the best parking spot to being in the first car to leave that parking lot at 5pm because they aren’t having to upload notes, fight with insurance claims. Imagine driving away with the realization you spent your day with your patient and not with Epic.
And finally…most importantly… Patients. Again and again, we had our audience and panelists to remind us of the whole point of it all…patients. With the complicated landscape of AI, technology, standards, policies, digital twins, and just plain cool tech…the questions we asked ourselves during the conference (and durinAI) …r existential insomnia triggered by high tech AI)… how do we make sure that we don’t become too dazzled by tech and focused on output that we forget about outcome and how do we ensure that our north star remains the patients? The answer lies in balance. As AI continues to advance and offer exciting possibilities, we must remain grounded in the core mission of healthcare: improving lives. It’s easy to get swept up in the potential of new technologies, but if they don’t fit into the current workstreams and if they don’t ultimately serve the needs of patients, then we’ve missed the mark. We need to ask ourselves tough questions at every step: Are we using AI to solve real problems or just creating new tools?
By keeping patients at the center of every conversation, every decision, and every innovation, we ensure that technology serves its true purpose: not to replace human care, but to enhance it. In the end, it’s not about what AI can do—it’s about how AI can help US provide better, more compassionate care. And that, more than any technological breakthrough, is what will define the future of healthcare.