Last week, I came across research on skill atrophy from overuse of automation. The article explained why the FAA is urging airlines to keep pilots proficient at manual flying through its Flightpath Management guidance. The overall message was simple. When automation carries everything, core skills fade. Like a muscle, if you don't use it, you lose it.
That article caught my attention, so I set a personal challenge for myself this past week. A friend was flying in, and I decided to drive to the airport to pick her up without GPS. No blue dot. No voice prompts.
On the beltway, my calm faded with every exit sign. A small surge of worry, then the next. Is my exit ahead? Did I pass it? Those feelings mattered. In that moment, I realized my navigation skills had gone soft. My little experiment validated what the aviation research warned about. Overreliance can dull fundamentals.
After confirming the atrophy of my personal navigation muscle, I decided I had to be proactive to maintain a healthy balance. In Waze, my GPS go-to, I found a setting for minimal prompts: fewer instructions, just key cues at the right moments. This new setting helps me stay both alert and calm. I am not outsourcing the whole drive, and I am not white knuckling it either.
As I continued to think about achieving a healthy balance with AI, it reminded me of lessons I learned growing up. When my mother thought I was working too much, she would say in Spanish,
Ni tanto que queme al santo, ni tanto que no lo alumbre.
Not so much that it burns the saint, not so little that it does not light him.
I once asked her what that meant. She reminded me of all the times she asked me to light the candle at our home altar. And she would always instruct me to place it close enough to light the saint, but not so close that it would burn or damage the statue. It was a delicate balance.
That picture of balance is how we should approach AI. Use automation just enough to enhance and amplify your capabilities, expanding the value you bring while keeping your skills sharp and your judgment in the loop.
Using AI is not an all-or-nothing proposition. Like placing a candle near a saint, you still use the flame, but at a safe distance so it illuminates without burning. Use AI to illuminate your work, not scorch it.
Myth #1: The more I use AI, the more advanced I am.
Reframe: More usage looks like progress, but quality drops when critical thinking fades.
What to do: Design prompts that force engagement. Ask the model to list assumptions, show each step, and cite lines from sources. Add one fast human check for logic and fit to context before anything ships.
Myth #2: Manual practice is a step backward.
Reframe: Polished outputs make practice feel unnecessary, yet short manual reps keep core skills and confidence ready for nonstandard cases.
What to do: Schedule brief manual reps each week. Rebuild one deliverable without AI. Compare to the AI version. Update your checklist of steps only you will verify.
This week, try the following tools and Prompts to Steal to help you take action. These will help stay ahead of the curve.
ChatGPT as a tutor
Prompts to steal
• “You are my tutor for [topic]. Start with five diagnostic questions. Do not give answers until I try. After each response, explain the reasoning and show the steps. Give a 1 to 10 confidence score. End with what I did well, what to practice next, and two stretch problems.”
Power tip: Turn on Study mode and learn if available. Ask for Socratic questions only on day one to force active thinking.
Perplexity as a debate partner
Prompts to steal:
• “Act as a debate coach. Present the strongest case for and against using AI for [workflow]. Cite two high-quality sources with dates. Ask me three questions that could change the recommendation. End with a reusable decision checklist.”
Power tip: Ask “Where could this reasoning fail” to surface blind spots.
Elicit for evidence checks
Prompts to steal:
• “Show me the top studies on [question]. Extract findings, methods, sample sizes, and limitations in a table. Flag weak designs. Tell me what new evidence would change the conclusion.”
Power tip: Click into method summaries to quickly identify shaky claims.
Consensus for research-backed answers
Prompts to steal:
• “What does peer-reviewed research say about [claim]. List consensus findings. Note disagreements. Provide short quotes with citations and publication dates.”
Power tip: Sort by study quality and recency. Cross-check any strong claim with at least one additional source.
Start with thinking, not answers. Ask the tool to outline a plan, list assumptions, show each step, cite source lines, and rate confidence. Save that prompt as a template. Compare the AI’s steps to your manual checklist. If they match, delegate more. If they drift, bring the work back to you.
Your move this week:
Use AI for speed. You set the standard.
Leaders who stay in the loop keep their edge!
¡Hasta la próxima, un abrazo fuerte!
(Until next week, a big hug!)
