The AI Revolution: It Still Needs To Start With The “Why”

Jason C. Bork, President and Founder, Pintail Solutions

Originally Published for Forbes Business Council

Traveling out of state to Rose-Hulman Institute of Technology as a freshman in 1991, I was not sure what to expect. In my first Calculus class, broad computer use was still relatively new. We had computer labs where we could complete assignments, but laptops had not yet been integrated into the classroom.

We did not use computers in that class. We did everything by hand. I knew how to “do” math. Hard math.

You can imagine my surprise when I walked into the joint final exam and realized that the other professors had taught Calculus differently. On that final, you, or at least I, could not simply “do” the hard math. The exam was designed to be completed using computers.

Over the next four sweat- and anxiety-inducing hours, several non-mathematical lessons became clear.

“Doing” the math no longer mattered in the same way. Computers could do it better and faster than I could. What mattered had changed.

First, understanding the real problem was critical. There was a lot of information, but what were we actually solving for?

Second, with so many numbers and so much data, it became essential to determine what was important. Once the problem was contextualized and framed, could I distinguish signal from noise?

Third, I had to recognize what was missing. From the problem, the equation setup and the available data, what was not there that still mattered? What needed to be inferred, assumed or tested?

Does this sound familiar?

Look at the application, and sometimes the misapplication, of AI. These are the same challenges we are experiencing today.

Understanding The Problem

I was at a recent life sciences luncheon discussing the impact and future of AI. It was a fascinating conversation, but the most interesting statement came from a successful serial entrepreneur who recommended that we stop talking about AI. He said this at an AI-focused event. And he was right. Instead, he encouraged attendees to talk about the problem they were solving, talk about the impact they were going to have and why it mattered.

AI is already having a significant impact on our work every day, and it will likely have an even greater impact tomorrow and the day after. But in many cases, people have lost sight of the problem they are solving and the benefit they are trying to creat

AI, LLMs and related tools are powerful means to an end. In life sciences, that end should be improved healthcare, better decisions, stronger execution and better outcomes. The tools are vehicles to deliver benefits and outcomes more effectively or efficiently.

Unfortunately, AI has become a buzzword that many people feel compelled to talk about. Organizations need to discuss these tools, but not without clarity about what they are solving, how they are being applied and why they matter.

More Data Is Sometimes Just More Data

The number of conversations I have had with frustrated potential customers across the life sciences ecosystem is staggering. Many share the same concern.

They hear the same desperate messaging from innovative technology companies: “But don’t you want more data?”

Then they tell me the companies selling AI solutions often are not sure what problem they are solving. Even more often, they are unclear about which data will truly help solve it.

I saw a similar issue years ago with Six Sigma. While working at a large pharmaceutical company, I was deeply pained to be part of yet another Six Sigma project.

Then, at the project kickoff, the Black Belt startled me by saying he had investigated the problem we were trying to solve and determined that three tools, out of the dozen possible, would provide both the data and the process needed to successfully implement the project.

I did not know they could do that.

The project proceeded at an accelerated pace, with focused commitment from the team. The lesson was simple: the value was not in using every tool. The value was in choosing the right tools for the right problem.

The same is true with AI. More data does not automatically mean better insight. More models do not automatically mean better decisions. Better outcomes come from understanding the problem, selecting the right data, applying the right tools and staying focused on the decision or action that must follow.

Bias Matters

As organizations build data processes and AI-enabled tools, what goes into the algorithms matters. What does not go into them matters just as much. We were working with an academic institution on early AI applications when a physician became confused and visibly frustrated. He tried to clarify what was going on and asked where we got our data. We explained that the data came from the EHR system and he replied that the data was inaccurate. 

There was a very long pause after that statement.

AI is powerful. Really powerful. But AI will not solve bad data or broken processes. In fact, it can magnify errors with great confidence.

Understanding the quality, depth, breadth and representativeness of your data is critical to successful AI implementation. Like clinical trials, if your data is not fully representative of your patient populations, you should not expect your models and algorithms to deliver reliable or equitable insight.

The same is true for operations, clinical development, healthcare informatics and business strategy. AI can accelerate work, surface patterns and support decision making, but it cannot replace the discipline required to understand the problem, question the data and interpret the output responsibly.

How many AI pitches truly understand your business and your needs? How many demonstrate a deep understanding of why the work matters?

The fundamentals have not changed. AI may change the speed and scale of how we work, but it does not eliminate the need for judgment. Organizations should stay relentlessly focused on outcomes that matter.