Artificial Intelligence: We’re All In
SriniVas Sadda, MD
President & Chief Scientific Officer
You may have heard about the poker game that was won this year by a computer equipped with artificial intelligence (“AI”). Sure, computers using AI had previously won chess games and Jeopardy! rounds, but poker is different. Poker requires not only calculating odds and good math skills, but also intuition and the ability to “read” your opponent. Evidence of that blend of machine learning and seemingly human intelligence in AI demands fresh thinking about how we want to integrate artificial intelligence into human life.
The real and exciting advances that AI offers in medicine matter because they mean we can more quickly diagnose, and precisely treat, our patients. I love that potential, and, even more, I love that our expanding understanding of the capacity and uses of AI is pushing us all to consider how and why doctors matter when intelligent machines can spot disease and recommend treatment.
Consider this example of how AI works in ophthalmology. To detect and diagnose eye disease, imaging is one of our primary tools. At our Doheny Image Reading Center, and other sophisticated facilities globally, the images are read by highly skilled experts. The truth is that impressive expertise can now be duplicated, or even bettered, by software that is trained to recognize (for example) telltale signs of macular degeneration. We teach the machine to mimic cognitive functions, in this case what patterns to look for, and the AI’s margin of error is extraordinarily low.
If we feed the computer enough data, it can effectively function on its own, making diagnoses and decisions. Mistakes can happen, of course, but with AI, mistakes can often be corrected, or avoided, by adding more information. In other words, there is no limit to the artificial brain’s capacity whereas at some point my own brain could be overwhelmed to the point of not being able to retain new knowledge.
All of this is prompting physicians to examine our role in medicine alongside artificial intelligence systems. What the machine can’t do seems obvious to me: people don’t want to hear worrying news about diagnosis or disease progression from a robot. A patient can’t be reassured by a machine. The difference at Doheny translates as care. AI can spot a problem and determine a treatment, but it can’t take care of you – a human touch is important.
The future of research and patient care is one of considered and careful integration. Pairing the best doctors with unbeatable artificial intelligence is the future. You can be assured our entire Doheny faculty is deeply engaged, asking good questions with a commitment to illuminating the way forward, and adding artificial intelligence to outstanding research and patient care.