Introduction
The fear of becoming obsolete is common, and it has a real basis. The way to stay confident when AI makes your skills feel outdated is to build evidence: run small experiments with AI on your real work, keep track of what your judgment caught and improved, and learn from people a few steps ahead of you.
Many professionals have had the moment. A task that took years to learn gets done by a tool in seconds, and the first thought is whether the rest of their skill set is next. Researchers and pollsters have started calling the feeling FOBO, the fear of becoming obsolete. It affects people who are performing well, which is part of what makes it hard to talk about.
The Fear of Becoming Obsolete Is Widespread, and It Has a Basis
The 2026 ETS Human Progress Report, based on a Harris Poll survey of more than 15,000 US adults, found that 58 percent of US workers report a fear of becoming obsolete. That figure rises to 74 percent among technology workers and 73 percent in financial services. More than half of workers (54 percent) say their priority has shifted from seeking job security to staying relevant.
The UK shows the same shift. Totaljobs research from spring 2026 found that 42 percent of workers worry their existing skills could become outdated, and described a move from job insecurity to skills insecurity.
Workers who feel the ground moving are describing something real. A 2026 survey from JFF found that only 7 percent of workers say AI isn't significantly changing the importance of any of their skills, down from 42 percent a year earlier. Peter Cappelli of the Wharton School adds that part of the anxiety comes from confusion: workers keep hearing they need new skills, but the specifics are rarely named.
Why Confidence Drops Even as People Use the Tools
Using AI more doesn't automatically make people feel better about it. ManpowerGroup's research found AI usage rose 13 percent in 2025 while workers' confidence in using the tools fell 18 percent, with the steepest drops among baby boomers and Gen X.
That pattern fits psychologist Albert Bandura's work on self-efficacy. As Greg Edwards of Missouri S&T puts it in summarizing that work, skill alone doesn't determine behavior; a person's belief in their ability to use the skill effectively matters as much or more. Bandura identified several sources of that belief, including mastery experiences (your own successes) and vicarious experiences (seeing people like you succeed).
Your Expertise Is What Makes AI Output Usable
In a 2025 study presented at the CHI conference, "The Impact of Generative AI on Critical Thinking," researchers at Microsoft Research and Carnegie Mellon University surveyed 319 knowledge workers who use generative AI at least weekly. Workers who had more confidence in the AI tended to think less critically about its output. Workers who had more confidence in their own abilities tended to think more critically.
The authors found that people often skip critical thinking when they lack the skills to inspect, improve and guide AI responses. They describe the work shifting toward task stewardship: verifying outputs, integrating them and overseeing the process, while responsibility stays with the person.
For someone whose skills feel outdated, that raises a more useful question: who can tell whether the AI did the task well? Your judgment, built over years of doing the work, is part of what turns AI output into something usable.
How to Build Confidence With Small, Real Experiments
Bandura's framework points to mastery experiences, and research on AI at work points the same way. Edwards, who studies technology self-efficacy, recommends job-specific training built on frequent, simple, low-stakes practice tied to actual work, and cites PwC's "prompting parties" as an example. A study of a month-long, self-directed generative AI program at a consulting firm found that employees' confidence, familiarity and frequency of use all improved, along with the complexity of what they used AI for. And a Harvard Business Publishing study found that AI-fluent respondents were twice as likely to say they learned through experimentation.
A practical version looks like this:
Pick one recurring task. Use AI for a first pass.
Compare it with your own judgment. Note what you corrected, added or improved.
Keep the record. Each entry is evidence of your expertise, and the list grows every week.
Learn from someone a few steps ahead. Juan Bendana suggests choosing one skill to sharpen over the next 90 days, spending the first 15 minutes on it today, and sending a message to someone further along. Seeing a peer succeed is the vicarious experience Bandura describes.
What Leaders Can Do
Workers mostly know they need to adapt; the barriers are practical. In the ETS survey, 68 percent of US workers said upskilling costs are hard to cover, 63 percent struggled to find time to learn while keeping up with their workload, and 57 percent said it was difficult to get employer support. Totaljobs found that 71 percent of UK workers would feel more secure if their employer invested more in training and reskilling.
Leaders can respond directly to those barriers:
Protect paid time for experimentation. With 63 percent of workers struggling to find the time, learning that has to squeeze in around a full workload tends to get pushed aside.
Name the skills that matter for each role. Specifics beat general calls to "upskill."
Make it safe to say "I don't know how to do this yet." People experiment more when admitting a gap carries no penalty.
Each of these also builds trust in leadership at a moment when it's under strain, which affects how willingly people take on change.
Where the Evidence Stops
Nobody can yet say which skills will hold their value five years from now, and Cappelli's point about vague advice applies to the research too. Most of the figures in this post come from surveys of how workers feel, several of them run by vendors or trade groups. The Microsoft and Carnegie Mellon study shows associations from self-reported data, and the training studies are small. The practical steps here are reasonable readings of that evidence, and the studies didn't test them directly. Confidence built through practice makes the change easier to navigate, and it can't guarantee job security; none of the sources claims it does.
Lead Your People Through the AI Shift With Confidence
Juan Bendana's Lead Through Change keynote, drawn from his work with more than 250,000 leaders, gives audiences practical tools to show up with confidence in an era of technological disruption and uncertainty. The Confidence Cycle helps people overcome imposter syndrome and build a growth mindset in the moments where it matters most. Both suit leadership conferences, employee development days and corporate events where teams are working through AI, and sales kick-offs where people are being asked to change how they work.
Book a call with Juan Bendana's team to give your people a confident way through the AI shift.
Confidence in the AI era gets built the same way it always has: with evidence, one piece at a time.