The AI butterfly effect: How a single stream of bad data can bring artificial intelligence models to a screeching halt

The butterfly effect is the idea that one small, seemingly trivial event may ultimately result in something with much larger consequences. 

Remember the movie Back to the Future? At the end of the film, Marty McFly was disappearing from the Polaroid picture because of the butterfly effect.  (When Marty decided to meet his mom and dad instead of biding time at Doc’s house, the future changed, causing a ripple effect that was the main plot of the movie.) 

In AI, the butterfly effect is a perfect analogy to explain how one single strand of bad data can cascade through an AI data model, leading to skewed analyses and misguided strategies. Before you know it, that erroneous steam of data has polluted the entire model, affecting its output over time. 

Mitigating this risk is one important centerpiece of an enterprise AI strategy. CEOs and CTOs, it’s time to go to AI school, starting with a core prerequisite. 

AI for Executive Leadership 

For business and organizational leaders, nothing is more important to make an informed business decision about AI than understanding what it is, its potential and its risks. The first thing you need to do is ask yourself a hard question: What exactly is it that you want AI to do for you, your organization, or your employees? 

In short, what core business problem are you going to ask AI to solve?

Do you need to help customer service agents do their jobs faster and better with automated case replies and case summarization as part of the case insights?  

Do you need to help marketers to work smarter and more efficiently with AI content creation and in-platform ideation? 

Do you need to help your developers code better with automated code suggestions and repository audits? 

Because most AI use cases at the start will be unique, each should be focused on as its own “thing.” After all, any enterprise AI program is going to be multifaceted and multilayered. And AI will affect various industries and organizational teams differently. Therefore, any enterprise AI strategy must start with a fundamental understanding that one size won’t fit all with AI. Far from it. It needs to be discussed in the context of your unique business and the industry you are in. 

The AI conversation is complex, technical, and scary. It’s also fascinating, interesting, and teeming with potential. CEOs, it’s up to you to set the tone that your teams will follow. 

After all, if you don’t lead your employees in the AI era, who will? 

---

Jennifer Johnson works in digital strategy at Horizontal Digital. She is based in Dallas.