You can learn a crucial lesson about working with AI by looking up. The sky holds a decades-old example of advanced automation. Modern pilots have flown millions of miles with sophisticated autopilot systems. Their experience offers a direct parallel to workers now handing tasks to AI. The key is not to avoid the tool. It is to master it strategically, ensuring your core skills remain sharp.
Steven Shaw is an assistant professor at King’s College London. He studies what happens when people outsource their thinking. He sees a clear link. “We’ve had autopilot in planes for a very, very long time,” Shaw noted at a recent summit. “So they learn to fly manually first and then they learn to use autopilot strategically.” This sequence matters. Pilots build foundational competence before they let the machine assist. Aviation authorities actively reinforce this practice. The Federal Aviation Administration has issued multiple alerts over the years. They warn that continuous use of autoflight systems can degrade a pilot’s ability to manually recover the aircraft. Their guidance is simple. Pilots should periodically switch off the autopilot. They should practice manual flying when conditions are safe. This maintains their essential reflexes and judgment. Shaw applies this directly to our new reality. “Even experts are at risk of deskilling,” he said. The danger is in repeatedly outsourcing your cognitive work without engaging your own mind.
This warning is gaining volume among leaders. OpenAI CEO Sam Altman has pointed to “cognitive atrophy” as a risk. French AI executive Arthur Mensch has cited “deskilling.” The concern is that overreliance on AI might weaken our underlying abilities. Early research suggests this is not a theoretical worry. It is already happening. AI researcher Vivienne Ming ran an experiment with 72 participants. She found that roughly 90 to 95 percent used AI either to generate answers or to validate their assumptions. They did not use it to challenge their thinking. “The overwhelming trend is substitution,” Ming concluded. Shaw’s own research involved over 1,300 people. His team found participants accepted AI recommendations nearly 80 percent of the time. This happened even when those recommendations were deliberately wrong. Shaw calls this automatic, uncritical acceptance “cognitive surrender.”
The path forward is not to reject AI. It is to change how you approach it. Your first step should always be your own. Shaw’s advice is straightforward. “Just think first and then go to the prompt,” he says. Starting with your own analysis widens your perspective. It prevents AI from narrowing your ideas from the outset. Other professionals model this approach. Geetha Rajan is a former AI adoption leader at PwC. She writes her own complete first drafts before using AI. She then employs the tool to check facts and pressure-test her arguments. Ryan Deiss, CEO of DigitalMarketer, advocates a clear framework. He suggests humans do the first 10 percent of a task. This means defining the problem and setting direction. AI can then handle the bulk of the execution, the middle 80 percent. Humans must return for the final 10 percent. This last stage is for applying judgment, refining outputs, and adding a human touch. This method ensures you are always the pilot, not just a passenger.
The lesson from the cockpit is timeless. Automation is a powerful ally. It excels at handling routine operations and complex calculations. Yet it works best under the command of a skilled human who has maintained their core competencies. You would not want a pilot who has only ever watched the autopilot fly. Do not become a professional who only ever watches AI think. Your goal is integration, not substitution. Build your skills first. Use the tool to augment them, not replace them. Periodically switch off the assistant. Practice your craft manually. This is how you stay in control. This is how you ensure that as technology advances, your own capabilities do not fade.
- Master AI strategically
- Build foundational competence
- Engage your own mind
- Conduct your own analysis first
- Maintain core skills
- Practice manual tasks periodically
| Expert | Advice |
|---|---|
| Steven Shaw | Think first before using AI |
| Geetha Rajan | Write first draft before using AI |
| Ryan Deiss | Do the first and last 10 percent of tasks |
