Since their public launch in late 2022 generative artificial intelligence (genAI) tools like ChatGPT have sparked huge excitement, along with worry, across all economic sectors. But genAI’s possible effect on health care seems significant. In a field where an estimated 30% of the $4.3 trillion spent each year in the US adds little to no value where many tens of thousands of people die from mistakes we can prevent.
The healthcare sector’s recent rush into digital transformation through the rollout of electronic health records (EHRs), mirrors the “productivity paradox” pattern.
Studies on the productivity paradox show two key reasons why technologies often fail to deliver their promised value. First early versions of many technologies have flaws; the tools that succeed are those that get better with each new version.
People can’t grasp or put into action the big changes needed in how organizations are set up, how they’re led, who works there, and how work gets done to make the most of new technologies, at least not right away.
Before we talk about the productivity paradox and its connection to genAI, let’s look at how this paradox has shown up with other health care technologies in the past. We’re talking about general purpose technologies here – the ones that have an impact on many different tasks and specialties, not the more specific ones like laparoscopic surgery or the tools used in interventional radiology, which are easier to put into action. In the last 15 years, the biggest general technological change in health care has been the rollout of EHRs (electronic health records).
While EHRs have reduced medication errors and brought many other advantages, their impact on productivity remains unclear when considering the added paperwork for clinicians. Doctors and nurses often point to EHRs as a main reason for their job frustration and high burnout rates.
Features of genAI that could speed up enhancements
To start with, genAI tools are super easy to use. While better prompts lead to better results, these tools don’t need much know-how from users. The crazy-fast adoption of ChatGPT and other language-based genAI (100 million users in just 2 months) is because you don’t need special training to use them.
Next, the fact that users can get genAI through software on their computers also speeds up how people start using it.
In the past, one of the main things that helped overcome the productivity paradox was how the tech got better through repeated improvements. A key feature of genAI is that it can get better over time without much human oversight.
In conclusion, the IT productivity paradox will show up again when we put genAI to work in medicine just like it did with other tech before, both in and out of healthcare. Healthcare has some unique features that make it even harder to get the promised benefits from tech tools than other industries.
Reference:
https://jamanetwork.com/journals/jama/fullarticle/2812615
Â
