Do you feel like AI has made your work faster but worse? You’re not alone.
For Josh Anderson, a software consultant with 25 years of coding experience, the first few weeks felt magical.
from From June to August, Anderson livestreamed on YouTube how he used AI to build software products without writing a single line of code himself.
The app, called Road Trip Ninja, was designed to help families find reliable stops on long drives – places with clean restrooms, decent food, and space for kids to run around. Users can record and rate locations along major routes, turning tried and tested road trips into something more predictable.
When we let AI build the app, functionality appeared within minutes. Progress was rapid.
But as the livestream progressed over several weeks and the codebase grew to about 100,000 lines, the pace slowed.
Interactions with the chatbot ranged from minutes to hours. The plan deviated from the standards he had set. And solving the problem became “a never-ending wrestling match,” he said.
By the time Anderson himself stepped in to make a change, something else was changing. It was his confidence.
Software consultant Josh Anderson. Provided by Josh Anderson
Anderson believed that if he needed to get behind the wheel during an experiment, he would get back to work without a second thought.
He has spent years building and maintaining systems to withstand real users and real disruptions. In fact, when he opened the cord, he hesitated.
“It wasn’t completely frozen,” he told Business Insider. “But there was hesitation in every move.”
This moment captures a risk that a growing group of workplace researchers say companies and workers aren’t paying enough attention to: AI is covertly unskilling people.
early warning signs
he is not alone. When Anthropic’s Claude passed away earlier this month, some developers said they struggled to keep their jobs. Tasks that used to be routine with AI suddenly seemed difficult without it.
“Claude’s outage hits even harder when you realize you’ve outsourced half your brain to him,” one Redditor posted. “You’ll be writing code like a caveman,” joked another.
Moments like this show that change is already underway. As AI increases production, it is quietly chipping away at the skills behind it.
John Nosta, founder of NostaLab, says that while AI can improve productivity, it can also quietly erode skills. Provided by John Nosta
John Nosta, founder of innovation and technology think tank Nosta Lab, calls this the “AI rebound effect.” This is where increased performance masks decreased capacity. “The skill set is actually below baseline,” he says. The danger is not just addiction, but regression.
Because AI systems provide quick and sophisticated answers, they can also distort the way people judge their own abilities. “AI is overestimating our capabilities,” Nosta said.
Part of the problem, Nosta said, is that AI reverses the way humans normally think. In traditional reasoning, people move from confusion to exploration to structure, and only then to confidence. AI reverses that order.
“The first thing we get to the answer is a reversal of the human cognitive process,” Nosta says.
If this reversal becomes the norm, the risks will be greater than the productivity. “Human cognition is on the verge of obsolescence,” he added.
the illusion of expertise
The risk of de-skilling is particularly acute for junior employees. Mascot/Getty Images
In the workplace, fluency is often mistaken for competence.
Rebecca Hines, director of Green’s Work AI Institute, said AI can create the illusion of expertise. She says it’s becoming increasingly difficult to tell where workers’ knowledge ends and where their skills end.
Her study details two potential outcomes.
When used intentionally, AI can create “cognitive benefits,” freeing up time and sharpening judgment, especially in areas where employees already have expertise.
When used reflexively as a shortcut, it creates “cognitive debt” that quietly erodes people’s skills while allowing them to run faster.
The difference is whether AI supports thinking or replaces it.
Drop your skills at the beginning of your career
Hines said the risk of de-skilling is particularly acute for early-career employees.
The junior role has traditionally been a training ground for learning how to solve sticky problems, fix what’s broken, and defend your ideas when someone challenges you.
Without that experience, employees can appear competent without developing actual expertise.
It may take years for the full impact of that change to be felt. But early signs are already visible, and those most at risk are those early in their careers.
“Most professionals today learned their craft before AI and have a baseline,” Jan Tegze, author of Job Search Guide and How to Talk to AI, told Business Insider. “The people who are at risk are those who don’t build that baseline at all.”
Some employees are recognized for how often they use AI tools Victoria Jones, Pennsylvania Images/Getty Images
Ben Eubanks, chief research officer at human capital advisory firm Lighthouse Research & Advisory, said there has always been a gap between the concepts learned in school and their application in the real world, but AI is widening that gap.
“You don’t have to mess around with problems, think of solutions, or challenge the way you’ve always done things,” he says. “Instead, you can ask AI a question and get a good answer instantly.”
This change is making it harder for younger employees to develop resilience, he added.
At the same time, some companies are starting to step up their actions.
Hines said employees are increasingly being measured on how often they use AI tools, with a focus on speed and results over deep understanding.
Ashley Hurd, CEO of Atlanta-based Manager Method, which provides hands-on training for leaders, said she has seen an increase in the past six months, especially at tech companies, to include the use of AI in performance reviews.
Mehdi Pallavi, chief executive of the digital economy think tank, said workers may need to use a “mental gym” to stay smart in the future. Courtesy of Mehdi Pallavi
But problems emerge when something breaks or when employees have to think through the problem themselves, said Sara Gutierrez, chief scientific officer at talent appraisal firm SHL.
When Anderson returned to work, he felt the change firsthand.
“I knew how it worked,” he said. “But I didn’t understand them [workout] The person in charge for those three months. ”
This is causing some leaders to rethink their training. Mehdi Pallavi, CEO of the International Data Centers Agency, a digital economy think tank, said companies may need “mental gyms,” or spaces where employees can intentionally practice problem-solving without AI, similar to using a gym to build muscle.
“My swing went crazy.”
Anderson’s experiment shows what happens when you don’t do these mental exercises.
By the end of the experiment, he was proud of what he had created. But when he himself returned to make changes, something felt off.
He compared it to watching someone play golf. You can study swing. You can understand how it works. You can say, “This is how it should be done.”
But that doesn’t mean your body knows how to do it. And when he finally got the club back in his hands, he felt it.
“The swing was crazy,” he said of writing his own code again. “I thought, ‘But I know how to do it,’ but I couldn’t move my body the way I wanted.”