The penguin problem is a common phenomenon explaining the adoption of technology or innovation. Different from the Gartner hype cycle or the Innovation adoption cycle, the Penguing Problem states that people will only adopt when everyone moves, so no one moves.
The Penguin Problem is closely associated with network effects. A platform or technology is only valuable to me when many people are also adopting the platform or technology. But someone has to take the first step.
“The Penguin Problem” — No one moves unless everyone moves, so no one moves.
In HealthCare, the adoption of electronic medical records (EMR) faced the Penguin Problem. In 2009, (Lynn, 2009) and (Kuraitis, 2009) portray the problem in the industry. Doctors were reluctant to utilise EMR as they did not perceive any additional benefits to their “current” ways of working. Today, EMR is the basis in which our medical data is archived.
The adoption of some technology is low risk: switching from Hyves to Facebook was free, and the platforms were not mutually exclusive. If Facebook did not live up to the hype, people could always go back to Hyves. For the adoption of EMR, implementation costs and learning costs were involved, but the electronic medical records could, temporarily, be run next to the paperwork in case of not living up to the expectations. Viteezy runs an algorithm, based on your answers to pre-existing questions, and gives recommendations to certain vitamins and supplements. Taking vitamins and supplements usually is low risk, but may be high reward.
But what happens when the adoption of a certain technology is high risk? What if certain data can be hacked or used for reasons that you are unaware of? What if the technology affects your health insufficiently or negatively? Artificial Intelligence in healthcare is upcoming, and the results are promising. AI can do small tasks, like automate reminders to take medication, or complex matters, like identifying people at high risk or personalizing dosages of medication. Many patients may be reluctant to leave complex tasks and decision-making to “robots”, even when the results are reviewed by a doctor.
Some may argue that human error is the greatest error, so that AI is more robust. But, the AI is made by human hands, exposing it to human error once more. However, the risk of human error in creating the AI is much smaller if we demand it is reviewed by multiple parties and thoroughly tested for errors. Furthermore, removing bias in decision-making is necessary to avoid the AI taking ethical decisions that the human brain cannot even process (like the “trolley problem”).
The adoption of technology innovation in healthcare will remain an intricate problem. Everyone believes the predictions of the innovation and the benefits they may bring, but no one wants to be the first. “Time shall heal all wounds” may be a cliché, but as time goes by and results remain positive and promising, some early adopters will help the group of penguins jump into the sea and reap the benefits.
Kuraitis, V. (2009, September 2). Overcoming The Penguin Problem: Setting Expectations for EHR Adoption. https://e-caremanagement.com/overcoming-the-penguin-problem-setting-expectations-for-ehr-adoption/
Lynn, J. (2009, October 23). Penguin Problem in EMR Adoption | Healthcare IT Today. https://www.healthcareittoday.com/2009/10/23/penguin-problem-in-emr-adoption/
Really interesting article. I had never thought about innovation through these lenses. Even though AI has been around for some time now, I think it’s still not really palpable for most people, so, using the situation described on the article: when it concerns their own health they can be more resistant to accept a “machine generated” diagnosis for example. However I do believe that once AI and similar emerging technologies become more widespread and people get more familiarized with it, these resistance barriers will get smaller and smaller, there’s still a lot to learn from AI, but in a few years and with more advancements, I believe it will become clear to the majority of the population how an AI analytics capacity is light years ahead of humans’.