AI can make workers better — then worse at their jobs, an innovation theorist warns

• John Nosta, founder of NostaLab, said AI can boost productivity while quietly eroding skills.
• He said that reliance on AI can cause abilities to fall below baseline once the tool is removed.
• The effect inflates confidence and creates long-term risks for workers and employers, he added.

AI is often sold to workers as a pure upgrade — a way to write faster, analyze better, and perform at a higher level with less effort.

However, John Nosta, an innovation theorist and founder of NostaLab, an innovation and tech think tank, said that framing overlooks a crucial downside: what happens after the boost.

In his view, AI doesn't just enhance performance; it can also weaken the underlying skills people rely on when the technology isn't there.

"The skill set actually falls below baseline," Nosta told Business Insider, describing what he calls an "AI rebound effect."

When AI makes you better — then worse

Nosta compared the effect to a doctor performing a colonoscopy with the aid of AI.

With AI scanning alongside the clinician to help spot small polyps, the doctor gets better at the task, he said. The problem arises the next day, when the same doctor performs the procedure without the aid of AI, he said.

"I have to go back to the regular way," Nosta said. "And the skill set actually falls below baseline."

The danger, he said, isn't just dependency — it's regression.

Confidence rises as competence slips

Nosta also warned that AI can distort how workers judge their own abilities — a concern shared by many academics and researchers, including Rebecca Hinds, head of the Work AI Institute and Nobel Prize-winning physicist Saul Perlmutter, who have said that AI gives the illusion of understanding, while weakening judgment.

"We actually have an overinflated sense of ability through AI," said Nosta, who described the effect as "really dangerous."

In his view, AI doesn't just help people do more. It makes them feel more capable — even when that confidence isn't backed by independent skill.

That false confidence can be risky, especially in high-stakes environments, he said, where workers may take on tasks or decisions that exceed their real judgment once AI support is removed.

A growing workplace risk

Nosta described what he sees as a growing "cognitive codependent relationship," especially among younger workers entering AI-saturated jobs.

Used deliberately, he believes AI "makes me smarter." Used as a substitute for thinking, he said, "it's going to make me dumber."

Researchers at Oxford University Press reached a similar conclusion in a report released last October, saying that AI makes students faster but less deep in their thinking. Other academics have taken it a step further.

Kimberley Hardcastle, a business and marketing professor at the UK's Northumbria University, told Business Insider last October that heavy reliance on AI can lead to the "atrophy of epistemic vigilance" — the ability to independently verify, challenge, and construct knowledge without the help of algorithms.

To avoid "cognitive atrophy," Nosta said, "we have to sustain a level of cognitive risk."

His prescription is intentional resistance: preserving "cognitive grit," maintaining friction, and using AI to learn rather than to bypass learning.

In the AI era, he added, the biggest threat to work may not be smarter machines — but humans slowly forgetting how to think without them.

"For the first time in history, human cognition is on the obsolescence chopping block," he said.