Is Chat GPT Code Red the Start of a New Epoch?

Artificial Intelligence (AI) has been capturing a massive amount of attention lately as corporations, lawyers, and individuals are wrestling to understand the short-term and long-termimpacts.  In recent weeks, Open AI made a splash with the release of Chat GPT.  Open AI is a research and development company created to further the advancement of AI for all humanity.  Usage of Chat GPT has been exponential as people are throwing all sorts of challenges at the product with curiosity to see the results.  The ability to create a unique song, a research paper, digital painting, or medical progress note in the blink of an eye is both fascinating and scary.  The industry is scrambling to respond to potential abuse (NYC banning Chat GPT in schools), defend their businesses (Bing & Google) and wonder how to harness.

AI is the blanket term that encompasses machine learning, which includes deep learning to show the progression of how machines can mimic human behavior.  British logician and computer pioneer Alan Turing spent considerable time defining and pushing the boundaries of computers to let the machine learn from experience. The Turing Test was created to determine if a machine had true “intelligence.”  The term AI was officially defined by Computer Scientist John McCarthy in 1956 as the science and engineering of making intelligent machines. These visionaries sparked the attention and interest of scientists, governments, and even movie makers as they unpacked the notion of what could happen if machines became intelligent and aware. The challenge over the last half-century has been balancing the ambition of our dreams with the reality of hardware limitations. The amount of knowledge in the human mind has been a major stumbling block in trying to capture and replicate viable models. 

Thus far, the output of Artificial Intelligence to most businesses has been limited to automating repetitive tasks (Robots), cognitive computing (computer models to mimic human thought), and some aspects of machine learning (cybersecurity and natural language processing).

As an individual, we have seen AI most noticeable in the form of chatbots, voice assistants, and recommendations on our shopping history.  Helpful, convenient, and fun, but AI hasn’t been life-altering….yet. 

“Thus far, the output of Artificial Intelligence to most businesses has been limited to automating repetitive tasks (Robots), cognitive computing (computer models to mimic human thought), and some aspects of machine learning (cybersecurity and natural language processing).”

As professionals in Healthcare IT and leaders of Digital Health and Transformation for our health system, we’re actively watching and pursuing how we can harness the benefits of AI to impact healthcare.  We have it segmented into “three logical groupings” of utilization, patient-facing, staff-facing, and clinical.  From a patient perspective, informational chatbots, symptom checkers, and triage tools are real-world solutions that are becoming more prevalent to provide frictionless access to healthcare.  For staff, the goal is to automate manual work processes. 

Activities like insurance verification, new staff onboarding, financial reconciliation, and supply chain procurement allow a stretched workforce to become more operationally efficient.  Given staffing shortages and financial resiliency, these projects have been prioritized. 

The clinical side is the most exciting and ripe for opportunity, as we’ve barely begun to scratch the surface.  I’ve heard AI thus far compared to an iceberg where we are just seeing patches of ice above the water.  Current capabilities are predictive models and natural language processing (dictation and scribing).  The real advances yet to come are in the way of clinical decision-making (through imaging, data aggregation, or connected devices), coding recommendations based on documentation, and treatment recommendations.  For so long, we’ve been pouring data into the proverbial black hole of systems like EHRs and ERPs.  The challenge now is there is so much data; we cannot quickly make decisions from all the data.  Enter the ideal balance with computing power being able to sort through existing data, trends, and global data and make intelligent recommendations.  With Epic, we’ve been collaborating on data with their global aggregator (Cosmos) for years now.  The repository of data in a few years will be massive. Add in connected devices, like wearables, and suddenly we’re staring at the firehouse of medical information overload.  Enter the raw processing power of computing to sift through the data, help draw logical conclusions, build synapses in seemingly disparate points, and start to help providers, payors, health systems, and patients proactively manage their health.  One of my healthcare mentors says the health system of today is repair medicine.  Imagine the day when we can truly be in the business of proactive medicine and health.

The question becomes, where will AI leave us as humans?  For the next decade or two, it’s likely still building the models. While we can have all the data, a key component of providing care is compassion and empathy.  Successfully building the soft skills needed to navigate human psychology and situational awareness requires a lot more work.  The migration from indexing knowledge to turning it into logical connections is going to be the magic that turns it from a parlor trick into massive enterprises.  You can see the top Fortune companies (Microsoft, Google, Amazon, Apple, Tesla, IBM, etc.) all pouring money into it to reap long-range impacts. 

Now that the data has been entered and technology is fast enough, we need to turn data into decision-making.  While we’re still a whileaway from Skynet and the system becoming self-aware, I wrestle with the question, “Does the technology make us stupider?”  My sense of direction is horrible, and I’m lucky I can get home without GPS.  I can’t sit through a movie without searching IMDB to see what else that actor appeared in.  Social conversations with friends often leave several of us searching for “what year did that happen in?”  If we punt intelligence to machines, where will that leave us? I’m not advocating the return of the wooden card catalog boxes at the library, but if my 9-year-old daughter can write a structured medical progress note for a 56-year-old patient with an ischemic stroke in 15 seconds, our education system and future role in work are poised for a massive shakeup.