AI IS NOT A MAGIC PILL: Explaining the Diagnosis of Artificial Intelligence Solutions

This November, Novatio Solutions was awarded Rising Star at the UiPath Forward event — making us the only company to receive awards from the top two RPA firms in the world. At that event, I shared the stage with leading automation experts from Deloitte and Capgemini to provide thought leadership on the future of process automation.

In that panel discussion, I compared AI to medicine. And how, as I branched out into my second career in using RPA to automating knowledge-based work, I couldn’t help but think back to my time as a physician and in what ways it is relatable to the AI phenomenon.

In both occurrences, there is never a magic pill to cure all sickness or solve all problems. Medical treatment plans are a collaboration between patients and doctors. Similarly, RPA solutions are not “one size fits all,” and they take significant work on both sides, beyond the flip of a switch. Let’s explore more deeply.

The Trainwreck Patient

In my years as a practicing doctor, I would regularly see patients with a long list of medical issues. These stereotypical “trainwreck patients” had not paid close attention to their health, and it had taken a toll on their bodies over the years. Diabetes, heart failure, obesity, high blood pressure — they suffered from numerous ailments from years of neglect but expected me to prescribe a special antidote to fix them all.

The corporate world has trainwreck patients, too. In companies, inefficiencies pile on top of each other, magnified by year after year of neglect or duct-taped solutions. Organizations often request a quick fix to instantly and comprehensively solve all of their process problems. Enter artificial intelligence, the powerful elixir to repair all broken systems and undo a history of bad habits.

In both circumstances, there are responsibilities on each side of the solution. Each stakeholder — from the consultant to the person in need — plays a part in successful implementation.

Here are the 5 biggest keys to winning RPA solutions (and how it compares to similar steps in effective medical treatments):

 

1. Confirm Expectations

With all the hype around artificial intelligence and process automation, one of the most important roles of firms like ours is to set expectations. We can’t just turn on a solution and let it seamlessly take over processes. There is a significant amount of assessment, prioritization, testing, and evaluation for both parties to perform.

In the same way, physicians must help their patients understand the steps and challenges of treatment. When treating multiple issues (like in those trainwreck patients), I had to communicate that all ailments will not be cured overnight. In medicine and process automation, you peel away layers of problems and prioritize solutions. For trainwreck patients, you must pinpoint which issue is the most harmful and fix it first. In many cases, one problem triggers a domino effect of complications.

With diabetes, diet and exercise are just as important as medication. Heart disease and blood pressure are controlled by modifying lifestyle as much as taking drugs. Time and effort are needed from the patient in order to see results. The same is true for corporate clients.

Selling a patient or client on a solution that requires no work on their part is neither honest, fair, or giving them the best chance to succeed.

 

2. Combat Misinformation

Because of our access to endless information, it becomes a challenge to fend off inaccuracy. The most frustrating cases as a physician came when patients would come in with a self-diagnosis (from WebMD or another online source) and tell me what to prescribe them. Valuable time and energy would be spent breaking down the self-diagnosis before we could get to a more informed one.

Companies do this, as well. Artificial intelligence is a hot topic in the technology sector, with tons of interest and even more press to encourage that interest. Executives and tech professionals can get sucked into the hype, without the benefit of experience or context. It’s both our jobs to bring an open mind — to the concerns of the solution seeker and the expertise of the problem-solver.

 

3. Cultivate Trust

The best way to approach treatment plans and automated solutions is to develop trust together. If I give a plan to a patient or client, we need to have an understanding. I trust they will follow their plan intently — medication dosage and frequency, lifestyle changes, diet discipline (or process updates and organization changes). They trust I have their best interest in mind and will steadily monitor their progress.

We work together to support the success of the solution.

 

4. Constantly Evaluate Results

A key to reinforce trust building is to avoid a “set it and forget it” mindset. In some cases, the first solution is not the best one. Regular audits at every step are needed to ensure a long-lasting answer.

When I prescribe medication to a patient with diabetes, it might not work (or may not work immediately). People respond uniquely to medications, so new approaches may be needed for each patient, even if symptoms are similar. If you’re going to fail, fail fast, learn, and take another approach.

With corporate clients, we continually reevaluate the path to success. Once we find a solution that works, we follow-up constantly. The last thing you want after fixing a problem is to revert back to old ways.

Like in medicine, the best treatments are often preventative — before a problem sets in. While a patient might get kidney screens to prevent disease or stay informed on new treatment options, an organization should perform routine appraisals of their development and explorations into new technology.

 

5. Coordinate Collaboration

Effective collaboration is the tie that binds it all. In medicine, physicians and patients must rely on an ecosystem of people, working together, to overcome ailments. It’s not all left up to a solitary surgeon. Every successful surgery depends on a system of flawless, open collaboration. Then they evaluate and repeat, again and again.

Organizations implementing automation are no different. Integral to an effective AI process is a collective support team.

Gokul N. Solai, MD.

Leave a Reply

Your email address will not be published. Required fields are marked *