How to use AI without getting dumber, according to a serial entrepreneur

• An entrepreneur says overuse of AI will dull workers' critical-thinking skills.
• And he has a rule for his team and kids to stave off AI atrophy.
• Start with 10% human thinking, let AI handle the middle 80%, and finish with 10% human judgment, he says.

AI acceleration doesn't scare Ryan Deiss, but AI atrophy does.

The founder and CEO of marketing firm DigitalMarketer and The Scalable Company says the real risk of artificial intelligence isn't that it will outpace humans but that humans will stop exercising their own thinking.

"Back in the day, people didn't need to exercise because the work itself kept us strong. Industrialization changed that. Now we need to work "out" because our actual work doesn't keep us fit," Deiss wrote in an X post on Monday.

Now, he suggested, the same technology that made modern life more sedentary could also make workers "stupider" if they rely on AI to do all their thinking.

"The technology that made us fat is going to make us dumb, too," he wrote.

His solution is what he calls the "10-80-10" rule: the first 10% of any task should come from original human thought, then AI can take over for the middle 80%.

The final 10% returns to humans for judgment, refinement, and what he calls "de-slopification" — a slang term that refers to low-quality, mass-produced content created by AI.

In calling for reclaiming and exercising our independent thinking, Deiss echoes a growing warning from AI researchers and several tech leaders: overreliance on AI could dull the very cognitive and independent thinking skills that make workers valuable.

While Mehdi Paryavi, CEO of the International Data Center Authority, has told Business Insider overrelying on AI may be quietly eroding workers' confidence in their job skills, Anastasia Berg, an assistant professor of philosophy at the University of California, Irvine, said it is most harmful for junior workers. These workers lean on AI from day one and never build the foundational knowledge required to understand what it's actually doing, she said.

Software developers experienced skill atrophy firsthand earlier this month when Anthropic's Claude AI models suffered several outages, some saying that the disruptions laid bare how reliant they had become on AI tools.

According to Deiss, the people and brands that maintain the discipline to think first — instead of reflexively prompting AI for answers — will have a structural advantage in an AI-powered world.

"The advantage goes to people who invest the time to think and process," he wrote.

"Because the people (and brands) who still CHOOSE to think will have all the advantage in an AI-powered world," he added.