Passing to AI vs. Piggybacking off AI: How to Know When You're Using AI Too Much

What is the difference between passing to AI and piggybacking off AI?

Passing to AI here does not mean passing off AI-generated work as your own. It means handing AI a task you have already learned to do.

This is Ryan Vet's framework for deciding how much of your work to hand to AI. It is not an infallible framework or a rubric for exactly when to start and stop using AI. It is a personal gut check, and it comes down to one question: are you passing or are you piggybacking off AI?

How do you know if you're passing or piggybacking? The take-back test

Could you take back over? If you can catch the ball back at any point, you're passing. Ask yourself:

If yes, you're passing. If no, you're piggybacking.

There is a second form of the test for leaders: could you take care of your own yard before you start on somebody else's?

Where passing stops

Passing stops when you go beyond the capability you would ever have interfaced with another human being for. Picture the right fielder in Little League who throws to second base without looking, and the second baseman isn't there. It's degrees of separation. If you would ordinarily have worked with a person who had that expertise, that's fine. The trouble starts when you pile layer on layer and you couldn't step in yourself.

When is it fine to use AI?

When you've earned your spot. In a company, nobody plays every position. A director of sales can still pass to the marketing team, and they grab the ball and run with it. The same is true with AI, as long as you're still in the game.

Here is how it works for me, Ryan Vet. I have the graphic design skills, the video editing skills and limited development skills. I used to run a creative agency, and I would sit over the designer's shoulder, the motion graphic artist's shoulder and the video editor's shoulder. So when I use AI for that work, I'm passing. I could pick up in Premiere at any point.

I want to be careful not to get self-righteous here. I have more experience than many people, and not everyone had the luxury of running an agency. The point isn't who has earned the right. The point is to know which one you're doing.

The gut check I use: are you using tools to augment your own ability in a way that is better for you and the people you serve? Or are you shifting all the effort and weight to AI for something you're actually incapable of doing on your own?

What are the two kinds of piggybacking?

Piggybacking is the kid on a parent's back. Rarely could I give my kids a piggyback ride before they could walk, because they didn't have the core strength to sit up on my shoulders or hang on. With AI, piggybacking shows up two ways.

1. 01

   You're unable right now. You lack the skill or experience, so you hitch a ride. There is nothing wrong with piggybacking for a season. We all have to learn. It becomes a problem when it becomes the norm, because you trade away the chance to learn and grow.

2. 02

   You have no desire to learn. You can't do the task, and you don't want to try. This is where the kid stops growing. If they always ride, they may never gain the strength they need.

Is piggybacking always bad?

No. I piggybacked my son through the Garden of the Gods on his first birthday. Does he remember it? No. But I believe some memory formed of being outside in nature, sitting on my shoulders.

In my kitchen, my kids watch me cook with sharp knives. They aren't allowed to touch one until I decide each is responsible, and then they move up to a plastic knife. My four-year-old just cut potatoes with a plastic knife for loaded baked potato soup. Not all of them, because that would have been inefficient, but he started. He watched first.

Kids learn the same way at soccer shots. Each week it's a new move, like a pullback or a step-over. They have to learn them, and that's fine. Put them in a game with players five years older and they can't keep up yet. Sometimes a piggyback ride is a step on the way to learning more.

The problem is when you're a full-grown adult still being carried.

Speed isn't the question either. My brother hiked the Appalachian Trail, about 2,200 miles on foot. He could have gone faster on a horse, in a car or on a plane. He chose to walk. What are you giving up when you ride?

What does piggybacking cost you?

• 01

  The skill itself. If you always ride, you may never gain the strength.

• 02

  The wisdom. I can have all the knowledge and information in the world, but if I haven't done it myself, that's not wisdom. I haven't internalized it yet.

• 03

  The pushback. When you pass to a teammate, they can yell at you, look disappointed when you hog the ball, and call you out when you weren't ready to catch it. When you piggyback on AI, you're no longer on a person's shoulders. You're at the mercy of a machine's algorithms.

• 04

  The friction. Most large language models are built to make you feel better about yourself. They tell you it was a great idea, or fall on their own sword when they hallucinate. That eliminates the friction of teamwork.

• 05

  The pressure to perform. We innately feel pressure to perform for others. Someone depends on you, and that creates a healthy kind of stress. AI removes it.

What do leaders lose when they pass to AI?

In Ryan Vet's view, leaders lose something when they pass to AI. You could argue that executives are just passing to AI what they would have passed to a young employee. But the risk is no longer just passing. The giver of the task also loses ability.

When a leader tells a younger employee how to create the "we-ness," one of my favorite lines from Friends, someone still has to describe how to do it. The employee interprets it and gives it back. There is a progression, a give and take, and little nuances of instant pushback before the result is even delivered. Hand that to AI and you lose them.

Leadership is a lot like prompt engineering. You know what motivates each person on your team and what demotivates them, and you custom-tailor your instruction and feedback to keep the team healthy. A web developer, a designer and an ads specialist each hear it differently. AI is one person doing everything, and it doesn't stretch you to think about how to relate to each person. What that give and take does to us is unmeasured, and we are losing it.

There is a generational wrinkle. Younger workers tend to follow granular, step-by-step instructions and are less practiced at filling in gaps. Many had parents who did their homework, drove them to interviews and filled out their college applications, so they never built those small, steady skills. Leaders who only pass to AI are building the same dependency.

Put two robots on a basketball court and they would always beat the humans. They'd know the exact trajectory, the weight of the ball and every angle. But who watches that game? There is something gained in the friction of poor instructions.

Where do you draw the line?

How much can you outsource to AI is an ongoing conversation that won't be resolved anytime soon. I recently saw a LinkedIn post condemning any thought leader who uses AI for writing, arguing that AI would inevitably influence their output and make it disingenuous. The next day I saw the opposite view on Facebook: executives and thought leaders have long passed their writing to ghostwriters, and some of those books became international bestsellers even though the leader never put pen to paper.

Which one is right? Is that passing or piggybacking, and what decides it? There is so much moral ambiguity that a clean line would require a full understanding of every person's capabilities and exactly how much or how little AI interfered. This framework won't draw it for you. It's a diagnostic for self-prescription and reflection.

Questions people ask

Take the test, cite this page, read more