Gen Z uses AI to learn the most, but 57% struggle to apply it at work, survey finds

But while a chatbot offers somewhere to take those questions—to explain an unfamiliar concept, walk through a spreadsheet formula, or rehearse interview answers over and over—when it’s time to execute its answers, AI might get it plain wrong. Take a new hire asking AI for an Excel formula to calculate sales growth. They might copy the formula and get the right answer without understanding why it divides the change using last year’s sales. So when the boss asks them to change the calculation or explain the result, these younger workers still need to understand the math behind it.

A new survey released Thursday by language-learning platform Preply found many young workers feeling stuck with AI use, struggling to put their learning into practice. Gen Z respondents were the most likely to use AI for learning, and also the most likely to say they struggled to apply what they learned or were told by the chatbot in real life. For employers, that gap raises a practical concern: whether workers are developing skills they can use on the job and what support they need to get there.

Preply surveyed more than 5,000 working professionals across nine countries who tried to develop a career-related skill in the previous year. Among Gen Z respondents, 92% used AI for learning. Yet more than half (57%) said they struggled to put what they learned into practice, compared with 40% of Gen X respondents.

The findings don’t explain why younger workers struggled more, or whether AI played a role. Respondents were asked about applying what they learned across different methods, not just AI. Still, younger workers were more likely to feel their efforts weren’t getting them anywhere. Nearly three-quarters (71%) of Gen Z respondents said they felt busy learning without improving, compared with 56% of Gen X respondents. 

Josh Crossick, Preply’s chief product officer, sees learning as a process that includes acquiring knowledge, practicing, and building confidence over time.

That doesn’t mean the technology isn’t helping. People are still “learning more than ever and have access to tools more powerful than they’ve ever had before,” Crossick told Fortune. “AI is powerful. And AI is able to support in that journey, but it’s far from the whole journey.”

Apratim Purakayastha, then general manager of talent development solutions at workplace training company Skillsoft, previously told Fortune that AI practice tools have limits: “But this does not replace coaching.”

Crossick offered an example from his own experience learning Spanish while living in Spain for six years. His tutor taught grammar and vocabulary, but they also rehearsed conversations he would need outside the lesson—including taking his children to the doctor. 

That meant practicing how to explain a problem and responding as the conversation changed. A mistake didn’t end the exchange altogether; he had to work through it. 

“With another person, you’re in the role play. You have to keep going,” Crossick said.

Practicing with a person feels riskier than practicing with a machine, he said, and that’s the point: “The confidence that you build through success also goes up.”

Learning more, but feeling stuck

AI can make practice material easier to find. Todd Hughes, who trains language tutors at Rosetta Stone, previously told Fortune that he asks ChatGPT to generate Swedish reading passages at his level so he can practice without searching for suitable texts.

“For me, the whole AI thing is like this huge library,” Hughes said.

But easier access to practice material doesn’t guarantee people feel like they are actually improving and absorbing what they are taking in. Across the survey of all working adults ages 18 to 55, 76% said they spent more time learning than they did two years ago, while 52% said that extra time did not always bring better results.

The subjects ranged from data science and AI to business management and public speaking. Improving performance in a current role was the most common motivation for learning, and 51% said their employer paid for some or all of it. That means companies are investing in learning that employees don’t always feel ready to bring to the office.

The coworker still matters

At work, practice and feedback can come through everyday conversations and interactions with colleagues. A colleague can explain why the team handles a customer a certain way, question an assumption, and give feedback after seeing how the worker actually performs. 

A chatbot can answer a question, but it can’t read the room. Hughes said real conversations depend on cues that a machine may miss. “I’m communicating with you because I see that your head is going up and down. You’re giving me some type of active listening. You’re following up on what I’m saying,” he said. “I don’t think that AI is there yet.”

Iain Smith, head of behavioral science at Sunny Workplace, a nonprofit employee engagement platform, previously told Fortune that new workers may miss opportunities to learn informally in a workplace where colleagues increasingly communicate through screens.

“They’re not necessarily now sitting next to someone who’s been doing the job for 20 years and learning the shortcuts and the hacks and the little ways of coping with a stressful day,” Smith said. “They’re turning up to Zoom.”

Smith also cautioned against sweeping claims about generations. A new employee’s experience of the workplace matters, and a generational label alone doesn’t explain how they learn or perform.

For employers investing in AI, Crossick said, providing the technology is only the beginning.

Workplaces are complex, Crossick said, and new hires need someone who can help them understand “the human side of the system,” a person to bounce ideas off who “has been there before and has seen how things work.”

“The tools are just the tools,” he said. “You need to combine that with coaching and support and an understanding of how to apply those tools.”