What A CMO Learned Using ChatGPT Daily For 2 Months

I selfishly avoided sharing this out of fear that excessive demand would kick me out of ChatGPT. Now that it has already happened, I might as well hold open the flood gates. Over the last 2 months, I’ve used ChatGPT almost daily for everything from random questions to journaling and professional work. Here are my rambling observations:?

1. ChatGPT vs. Google: For anything that requires an internet connection or references recent events, I’m forced to go to Google. For pretty much everything else, I prefer ChatGPT. Quick access to information without sifting through Google results of ad-filled websites is refreshing and more efficient.?
2. Answers from ChatGPT feel like singular truth. This is advantageous for a simple and effective user experience but is a quagmire for bias, loss of knowledge, and omission of diverse perspectives. In the past, history was written by victors. I’m now convinced that in the future, history will be written by AI developers. It won’t just be history, it'll also be culture and sentiment of the populace. The importance of DE&I in the development and application of AI cannot be overemphasized.?
3. My favorite features based on my simplistic use cases are creating tables and searching for and synthesizing scientific articles up to year 2021 (pubmed replacement). I'm so basic.
4. While I abhor ghostwriting, I submitted some of my writing prompts for deliverables out of curiosity. The pleasant surprise was that the articles written by ChatGPT were really good - pithy, engaging, and easy to read. The unpleasant surprise was that I got so anchored to the tone and content of the generated work that I had a hard time finding my own spin on the topic. ChatGPT can stunt my creativity.?
5. I subconsciously try to be nice to ChatGPT. I use “please” and “thank you” and tend to phrase my asks as requests rather than demands. No, I don’t believe ChatGPT has feelings but its impressive humanistic qualities compel me to treat it well. Humans are wired for social connection and I believe we will inevitably project that onto the technology we create. Remember Tamagotchis? Even those rudimentary digipets brought my “friends” to tears when they died.?
6. Despite the above, using ChatGPT for “bot therapy” doesn’t work for me. The responses are empathetic and provide gentle guidance, but ChatGPT doesn’t know me. It doesn’t know my life experiences or thought patterns. It can’t pick up on subtle clues in my tone or word choice that reflect my subconscious. Once it can, it’ll probably be my new best friend.?

Now that everyone and their grandmothers are talking about ChatGPT, let’s talk more about AI and health. The applications of AI in healthcare to date have been largely focused on predictive analytics, precision therapeutics, drug discovery and development, clinical decision support, and imaging analysis. The repeating chorus line is that AI can reduce healthcare expenditure, increase operational efficiency (i.e. profits), prevent poor (i.e. expensive) health outcomes, and save some lives! These are value props for payers, hospitals, provider systems, and pharma. After all, they are the ones with the deep pockets. Partnerships with these entities not only bring in millions of dollars but also splashy press releases and credibility. Even the discussions on saving lives are never purely about the number of lives or improved quality of life. There is always a subtext of ROI.?

I fear that the greatest value AI can create for health (not healthcare) will be overlooked and never fully realized. What about the value prop for you, me, and Uncle Roger? What’s important to me isn’t knowing that I have a 2% chance of a heart attack in the next 30 days, or what autoimmune medication I should choose based on my genetics, or which providers are best matched to treat me. What’s important to me is knowing what I can do today so I never have to worry about a heart attack, how to prevent psoriasis flares so that I don’t have to be on medication, and what risks I should be asking my provider about that neither of us are aware of. I am an N of 1 and I can generate massive amounts of personal health data but I don’t know what’s signal from noise or what’s actionable. I’m my own black box and you are too.?

AI can solve for that. AI can be the ultimate tool to empower all individuals to live better lives and make better health decisions, not just healthcare decisions. It can be a savior and equalizer for some of the most challenging and inequitable issues we face in health. But will it??

Preventive health has been the lowest on the healthcare totem pole because the ROI isn’t always clear and usually isn’t realized in a year or two but rather decades. By that time, the individual is with a different employer, on a different health plan, in a different healthcare system. Who is going to pay? Who gets the ROI? How much ROI? If these questions remain at the forefront of how we evaluate new technology, we will continue to evolve but not leapfrog. Given the rising prevalence of chronic diseases, mental health crisis, and emerging conditions like long COVID; waiting for incremental change isn’t good enough. We need to leapfrog.?

If you’ve read this far, thank you. If you want to brainstorm a solution around this or if you are working on this and feel that I can help in some way, email me at [email protected].

#chatgpt #ai #health #healthcare #healthcareinnovation #healthtech