There is a flip side to the “don’t be left behind” rhetoric of technology innovations. Once a tech concept gets big enough to reach the zeitgeist, people will try to sell it to businesses indiscriminately. You can’t always assume that the one selling is concerned with whether the one buying actually needs it.

One of the exciting aspects of January’s PPAI Expo Conference was the massive crowds that showed up for the day’s great sessions on automation and artificial intelligence. These speakers drew standing-room-only audiences. I was particularly intrigued by the presentation of Eric Granata, CAS, founder of business service provider PromoPilot, which focuses on AI solutions.

In his session, “Streamlining Success: Automation Solutions for Distributors,” Granata touched on four things that shouldn’t be automated:

  • Tasks requiring human judgment.
  • Low-volume tasks.
  • Unpredictable tasks.
  • Problems that don’t exist.

In my role as business systems analyst lead at PPAI, I am well versed in the ways that automation can and has saved us time as an industry. We’re likely in the beginning of realizing those possibilities. But innovation requires communication, and there are only so many steps you want to skip in communication before you have created problems that will cost more time and money to solve.

Let’s focus on those first two points: “Tasks requiring human judgment” and “Low-volume tasks.” One particular ad I saw during the Super Bowl stuck with me. It involved two co-workers wanting to communicate something with their manager and being stressed about the correct wording. They opted to let AI send the email to their manager.

I can’t say I recommend that practice. We might not all have the same mastery over the English language, but often that lack of confidence in sending an email is based on a lack of clarity in a project. A supervisor needs to be able to read that confusion and act accordingly, often following up with the sender.

Using AI to polish something that is inherently flawed is a dangerous game. Or, in some cases, it may muddle a good idea that can still be conveyed – perhaps imperfectly – by the person behind the idea.

Some of you will read this and think “I would never do that,” but shoehorning AI into areas of work it cannot yet handle is becoming normalized. Some businesses will suffer from it. 

Tanaisha Dunbarger is PPAI’s business systems analyst lead