When an AI tool Jacob Adamson uses for work recently froze, so did his fingertips. The senior software engineer had become so reliant on the technology that he briefly forgot how to complete a code-writing task on his own.
“I felt the rust, as if I had come back to this code after a couple of days,” said Adamson.
He’s now concerned the same dependence could grip the five engineers he manages at Varonis, a data-security company. To keep their skills sharp, he said he may have his team do drills in which they can only write code themselves, without the aid of AI.
“We’re good at our jobs, but this technology can unknowingly lull us into this sense of dependence,” said Adamson.
AI tools are becoming ubiquitous across industries, helping workers interpret data, summarize notes, and crank out code in a fraction of the time it once took. Yet even as leaders tout the promise of AI, some fear there’s a hidden cost — a gradual siphoning of workplace know-how.
Underscoring the paradox is a new study from the University of Pennsylvania’s Wharton School of Business, which is based on a survey of nearly 800 decision-makers at US companies with more than 1,000 employees and over $50 million in annual revenue.
Nearly three-quarters of the respondents credited AI for delivering efficiency gains, while 43% said the tools may cause skill atrophy, meaning workers could lose their ability to perform essential tasks.
“There’s tension among leaders around whether AI is becoming a crutch,” said Jeremy Korst, a partner with consulting firm GBK Collective, which produced the study with Wharton.
‘It might make us dumb’
Sandor Nyako, a manager of about 50 software engineers at a large technology company, likes that AI tools make it possible to get work done faster. But he doesn’t want his team to lean on them too much for tasks that require critical thinking or problem-solving.
“If we develop an over-reliance on AI, it might make us dumb,” he said. “Someone would plateau at their current level.”
Nyako sees workers’ ability to think independently and solve problems on their own as essential to human evolution and innovation.
“To grow skills, people need to go through hardship. They need to develop the muscle to think through problems,” he said. “How would someone question if AI is accurate if they don’t have critical thinking?”
From horses to cars
Some AI proponents argue that tech-driven skill atrophy isn’t necessarily a bad thing.
“Very few people know how to ride a horse these days, but they seem to be able to get around just fine,” said Phil Gilbert, former head of design at IBM.
Related stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
By using AI tools, he expects workers to achieve the same goals, only faster and with potentially better results.
“Outcomes are what are important, not the inputs,” said Gilbert.
He added that as long as workers understand the nuts and bolts of their jobs, there’s no harm in them using AI to get ahead.
“You need rudimentary spelling and grammar skills to communicate in writing, but that doesn’t mean you can’t ever use a dictionary or spell-check,” said Gilbert.
New skills, old skills
Employees may need to develop new expertise to leverage the benefits of AI, such as the ability to write effective prompts, said Bob Chapman, chairman and former CEO of Barry-Wehmiller, a global manufacturing firm.
“How to use AI should be the skill we teach,” he said, adding that some forms of traditional education may not matter as much going forward. “I don’t remember my chemistry class in high school.”
Other leaders, however, are worried about what will happen if people don’t learn the basics. At some point, there will be workers who’ve only lived in a world with AI at their disposal.
“They are going to be lacking some of the skills that I had to learn when I was a junior engineer,” said Adamson, the manager whose coding skills got rusty.
That’s a problem, he said, because AI tools don’t always deliver accurate results.
“AI may eventually get there,” he said, “but it’s definitely not there yet.”
