The Revolution That Took Thirty Years, and Is Just Getting Started
- Dr. Alan Greene
- 3 days ago
- 8 min read
This piece is adapted from a keynote I gave at Stanford's Center for AI in Medicine Annual Summit:. The full talk is embedded below, and the written version follows.
In 1995, if I wanted to answer a patient’s question and the answer was not in a journal I happened to subscribe to, I got in my car. I drove to a medical library, pulled the Index Medicus off the shelf, looked up what had been published, then walked into the stacks and photocopied the article. That was the fast way.
My wife Cheryl was sick around that time and wanted to read the same articles. She was not allowed in the library. Patients were not permitted access to the journals. I snuck her in more than once.
It is worth sitting with how strange that world was, because we have almost entirely forgotten it. There was no Google. There was no PubMed database the public could search. There was no broadband and no Wi-Fi. You dialed a phone, listened to a modem screech, and waited a very long time for a single page to arrive. And patients were not routinely given their own lab values. The convention was to say we would call if something looked wrong. The Pew Research Center later found that when practices worked that way, roughly one in seven actionable abnormal results was never reported or acted on at all. The person with the most at stake was the one person who could not look.
The problem that started it
I was a young pediatrician in San Mateo, seeing about three patients an hour and having the time of my life. Then managed care came through the Peninsula and we were told to see four an hour. Then five. Some days I saw fifty children. My families were kind about it, but they were honest. Your hand is on the doorknob the whole visit. The appointments are too short. We are not getting our questions answered.
I told a colleague I wanted to hold evening meetings with the families and work out a solution together. He said it was a terrible idea, that asking would only convince them there was a problem and that whatever they suggested would be impractical anyway. So I started holding the meetings quietly, in my office, after hours.
Their first ideas were indeed impractical. Work an extra two hours a day. Work six days instead of five. But out of those conversations came a different one. What if you answered one question a day in writing, and put it somewhere we could all read it?
Cheryl taught herself HTML and built the site. Every evening she printed the questions that had come in. I picked the one that seemed most useful and wrote a page or two, not just what to do but why, the physiology underneath it, the way you would teach a medical student. Our entire marketing effort was two business cards sitting in two exam rooms.
When we surveyed the families a few months later, they told me they were glad I had gone back to longer visits. I had not. The visits had actually gotten shorter. What had changed was the number of touch points. Instead of seeing a family two or three times a year for a few rushed minutes, we were now in contact weekly or daily. When a child came in with an ear infection, the parents already understood what an ear infection is and how it forms. So the visit could be about their child rather than about the definition of the disease.
The moment it stopped being a website
Questions started arriving from families in other practices. That made me uncomfortable, and we wrote back telling people to ask their own doctor. They wrote back saying their doctor was too busy. Then someone posed as one of my patients. The message said a dog had bitten their child, and I knew that family did not have a dog. Someone was willing to lie to a physician simply to have a question answered.
That week I closed my practice to new patients so no one could accuse me of poaching, and we opened the site to anyone. Within twelve weeks a quarter of a million people from six continents had used a site we had never advertised. One message I have never forgotten came from a parent in India who had walked for three days to reach a computer, worried about a child, asking politely whether we might be kind enough to choose their question. They said they would wait for the answer.
None of that was about a website. It was about how badly people wanted to understand their own health, and how completely they had been shut out of it.
Screenside manner
I assumed that answering strangers would feel mechanical. It did not, because we developed a practice we came to call screenside manner. The first principle was simple. Before answering, ask yourself what the question behind the question is, then test it out loud. A parent writes in worried about a fever. Are you worried this is dangerous? Are you worried your child is uncomfortable? Are you worried you have done something wrong? Either they say, how did you know, or they say, no, not at all. Either answer brings you closer. I am still in touch with people from those years whom I have never met in person.
What the profession got wrong, twice
The Institute of Medicine convened the senior figures in American medicine to decide what to do about health information on the internet. The purpose of the meeting was to agree on how to urgently warn every patient never to look online, and to ask every physician to tell their patients the same. I was easily the most junior person in the room, and I stood up and said that within a few years every patient would look online before speaking to a doctor, that diagnosis and treatment would live there, and that our real work was not to forbid it but to make it as good as it could possibly be.
That instinct to withhold is very old. Most people believe the central promise of the Hippocratic Oath is to do no harm. The part you actually swear is to teach the art of medicine to other physicians and to their sons, and to none others. Our founding document is a guild document. It held for two thousand years, mostly unconsciously.
In 2000 we proposed a millennium version at the first e-health conference in Europe. Same length, largely the same language, one clause turned around. By precept, by lecture, and by every other mode of instruction, I will impart knowledge of health to everyone, everywhere. The claim was that we would not be diminished by giving knowledge away. We would be more valuable. Thirty years on I am confident that is true.
That work became the Society for Participatory Medicine, and the argument that patients should be drivers of their health rather than passengers in it. It took five years of effort before it became law, in 2014, that a patient is entitled to see their own lab results. The studies that came out of that period were consistent. When people genuinely participate, outcomes are better, costs are lower, adherence to a shared plan is higher, and both patients and clinicians are more satisfied. Participation itself turns out to be therapeutic.
The second revolution
We are now in a second wave, and it is larger than the first. The internet gave people access to information. Artificial intelligence, if we handle it well, gives people access to understanding.
Let me make that concrete. Cheryl was diagnosed with metastatic inflammatory breast cancer the day after I told her I could not imagine being happier. We were told she would not see the end of that year. She is here, thirty years later, and she owns her health more completely than anyone I know. Recently she loaded eighteen years of her records, labs, and imaging into a compliant AI system and asked what the three most actionable things for her health would be. What came back was, frankly, excellent. It caught subtle details I would have wanted caught, and built a ninety day plan. I had never done that exercise for her. No physician ever had.
Then she did the more important half. The plan did not fit her values, so she went to a second system, worked out what her actual priorities were, brought the plan back, and had it rebuilt around them. That is the whole point. The machine assembled the picture. She decided what the picture was for.
For thirty years I have watched people try to assemble the puzzle of their own health and fail, because the pieces sit in different offices, different labs, different apps. For the first time, a whole picture is possible, and billions of people can carry something resembling a world class medical educator in a pocket.
The thing I actually worry about
My fear is not that physicians will be replaced. My fear is passivity. Deskilling is real. When calculators arrived, basic arithmetic scores fell, even as mathematics advanced. The same can happen here, to patients and to doctors, if we are not deliberate.
So the purpose of AI in medicine should not be to build more capable machines. It should be to build more capable humans. More capable patients, more capable families, more capable caregivers, more capable physicians. If we get that goal right, the products tend to follow.
There is a related point about trust. Many people assume safety and innovation are in tension. I think the opposite. Trust requires safety, and trust is what allows innovation to move quickly. In the late nineties, when the early health sites were called in and told to regulate themselves or be regulated, a dozen competitors with completely different business models sat down and hammered out fourteen commitments on credibility, transparency, and privacy. Those commitments were endorsed by federal agencies, by the Federation of State Medical Boards, and by the World Health Organization, and they raised the floor for everyone. Something similar is needed now.
What becomes scarce
When one thing becomes ubiquitous, something else becomes valuable. Information is now abundant. Expertise itself is becoming abundant. What grows scarce is judgment, context, wisdom, and knowing a particular person over time.
I saw this clearly at a conference where an AI system reproduced a musician’s voice so accurately that the room was stunned. Then the musician walked out and sang, and the room came to its feet. Something moved through that space that the imitation could not touch. She was real. She had been born, and she would die, and she was there.
That is the argument for the kind of practice we have built at Crescendo MD. Not that technology is a threat to the relationship, but that the relationship is exactly what becomes most valuable as everything else becomes cheap. A concierge primary care practice that is deliberately small can do the one thing abundance cannot manufacture, which is to know you. Our physicians have the time to sit with what you have already read, argue with it where it deserves arguing with, and help you turn a great deal of information into a decision that actually fits your life.
The challenge in front of us is the same one that was in front of us in 1995. It is happening either way. The only question is how well we do it.
Dr. Alan Greene is a co-founder and Chief Medical Officer of Crescendo MD in Portola Valley, California, and a board-certified pediatrician who has spent more than three decades working on how technology can deepen rather than dilute the relationship between families and their physicians. He founded drgreene.com in 1995, one of the first health information sites on the internet, and helped found the Society for Participatory Medicine. He cares for families across Portola Valley, Atherton, Woodside, Menlo Park, Palo Alto, and the greater San Francisco Peninsula.
Comments