The AI Workplace: New Research on Employee Adoption

AI is happening, but not to us—or so we think. Irrational Labs surveyed 767 knowledge workers and found a telling pattern: 

• When asked about their own job, only 8% of people believe AI will replace them
• When asked about their colleagues’ jobs, 14% believe AI will replace their peers
• And most tellingly, 29% believe AI will replace workers in other industries

Said differently, we think AI is 3X more likely to replace other people’s jobs than our own. This is mathematically impossible. Why the disconnect?

The explanation lies in two cognitive biases:

First, optimism bias: our tendency to believe positive future outcomes are more likely to happen to us, while negative future outcomes are less likely. As Tali Sharot notes in her seminal paper, “We underrate our chances of getting divorced, being in a car accident, or suffering from cancer.” Through this lens, we all see ourselves as bulletproof.

Second, the better-than-average effect: We overestimate our own abilities compared to others. It’s why most of us consider ourselves above-average drivers or funnier than average. While optimism bias focuses on future outcomes, the better-than-average effect centers on our self-perceived traits relative to others. These biases can intersect—for instance, when people believe they’re better positioned than others to succeed or avoid harm (e.g., to become ill).

For organizational leaders preparing for AI integration, this means your employees likely underestimate their need to adapt. This isn’t because they’re in denial—it’s because they’re human.

This is one key takeaway from our 2025 report on The AI Workplace.  Our research with 767 knowledge workers reveals how these perception gaps shape workplace AI adoption. Below, we explore what organizations can do to drive successful change.

🖱️ CLICK BELOW TO DOWNLOAD THIS REPORT AS A SHAREABLE PDF ⬇️

Our research strategy

We sought to understand employee AI usage patterns and attitudes toward AI change. We surveyed 767 people who use computers for more than 50% of their workday, asking them:

• How they use AI now
• How their coworkers are using AI
• How they think AI will affect their work in the future

Key findings

People are adopting AI—that’s the straightforward part. The complex part is how this adoption is actually spreading through organizations. Our research uncovered patterns in who’s using AI, when they start, and what stands in the way.

AI usage and adoption patterns

37% of people are using AI regularly in their workplace: For a new technology, this is rapid adoption. Even more notably, 68% reported engaging with AI at some level in their work. People roughly divided into thirds: one-third never use it, one-third use it occasionally, and one-third use it regularly.

Networks drive adoption.  You are who you surround yourself with.  

Our survey data suggests behavioral contagion plays a key role in AI adoption. We’re more likely to do something when we see others doing it.  87% of participants know at least 1 person who uses AI. That group is 3x more likely to have used AI themselves in the past week than those who don’t know anyone who uses AI. In addition if you know MORE people using AI, this makes you more likely to use it yourself. 

Interestingly, behavioral contagion can also occur solely based on what we perceive others are doing, rather than their actual behavior. In our study this seems to be the case – respondents believe others are using AI more than themselves — 28% of respondents think that people are using it daily. But in fact, only 15% report daily use.

This belief drives adoption. Think about it. Would you use AI if you thought no one else was? Probably not. But if you believe 99% of your peers are using it, you’ll likely feel pressure to try it. Just as seeing neighbors install solar panels or friends using Venmo can prompt us to follow suit, observing others adopt AI increases our likelihood of trying it.

Implication: Make AI usage visible to drive adoption

Leaders wanting to encourage AI adoption should make AI usage more visible within their organization. Behavioral contagion works through awareness.

Consider:

• If Jeff is a master at optimizing workflows using AI, call that out in your next all hands meeting. Have Jeff demo his workflows.
• Create a Slack channel where people can post AI tool tips and tricks.
• Shine a light on yourself. Show how you use AI in your own work.

Making AI usage a visible social norm accelerates adoption.

People adopt behaviors they observe in others, and no one wants to fall behind.

Who’s using AI: The numbers

Senior leaders: AI usage increases with organizational rank. Our survey found 50% of entry-level and junior employees used AI in the last week, compared to 70% of mid-level management and senior leaders.

Men: Gender has historically predicted technology adoption—from blockchain and crypto to robo-advising and smart home devices—and AI follows this pattern. Male respondents reported higher AI use (66.8%) than female respondents (53.3%). Surprisingly, factors like age, race, and income showed no correlation with AI use.

Managerial support: The strongest predictor of AI use isn’t gender or title—it’s managerial support. When managers endorse AI, usage reaches 79%. Without that support, it drops to 34.4%. Those uncertain about their manager’s stance fall in the middle at 42.1%. While authority bias may play a role in this, corporate policy is likely the driving force—restricted access to AI tools naturally limits workplace usage.

Implication: Remove structural barriers to AI adoption.

We often try to solve a behavior change challenge through persuasion. But often what’s preventing change isn’t individual attitudes—it’s structural barriers. 

If a junior employee isn’t using AI, is it because they don’t want to try ChatGPT? Probably not. They might simply lack the time to learn new tools, or don’t have approval to expense them. The barrier isn’t attitude—it’s structure.

Successfully driving AI adoption requires coordinated effort across procurement, HR, security, legal, and IT departments to enable access. Managers and senior leaders can support this by rolling up their sleeves and systematically removing barriers to access for their teams. It’s not glamorous work, but it drives change.

When it comes to the gender gap, awareness is the first step. It could be related to job roles or men’s tendency to show higher risk tolerance. Regardless, understanding that women may approach new tools more cautiously can help your teams design more inclusive and effective adoption programs. The gaming industry offers a model—women’s use has risen in the last decade partly through specific efforts to make games more “female-friendly”.

The opportunity for adopting AI in the workplace

Most people see AI’s potential to help their work. 68% believe it can improve their job performance. Only 21% view AI as irrelevant.

Yet despite this positive outlook, only 37% use AI regularly. This represents an intention-action gap—a seminal concept in behavioral science describing the disconnect between knowing what we should do and actually doing it. Even when people understand something is true or important, this knowledge alone often isn’t enough to change their behavior. To actually change behavior,  we have to remove barriers to usage.

People show an interesting pattern when considering AI’s impact on jobs. Only 7.8% believe AI will replace their own job, but this nearly doubles to 14% when considering their colleagues’ positions.

The number rises even further when considering people in different fields—respondents believe those outside their industry are three times more likely than themselves to be replaced by AI.

This mindset—that we’re personally immune to replacement—has implications for how organizations and individuals prepare for AI. Leaders need to recognize they’re working against deeply ingrained human psychology: optimism bias and better-than-average thinking. Getting this right means addressing core psychology, not just technology.

Implication: Experience with AI reduces optimism bias and closes the intention-action gap.

Interestingly, participants in our study who used AI regularly were more concerned about job displacement than those who didn’t use AI regularly. Rather than suggesting ignorance is bliss, this points to an advantage: hands-on AI experience provides a clearer sense of its transformative potential. 

For managers, this points to a clear strategy: lead systematic initiatives to remove barriers to AI use. Don’t just share information about AI’s benefits—information rarely changes behavior. Instead, focus on making AI fundamentally accessible and integrated into daily workflows and tech stacks.

How scared should we be?

Technology has repeatedly transformed work rather than eliminating it entirely. When computers entered offices in the 1980s, secretarial roles evolved from typing to managing complex digital workflows. Spreadsheets didn’t replace accountants—they made accounting more sophisticated. 

But while society adapts, individual transitions can be harsh. 

• The rise of automobiles eliminated 99% of carriage manufacturers between 1890 and 1920. 
• The introduction of ATMs reduced bank tellers from 21 to 13 per branch.
• Manufacturing automation contributed to 7.5 million lost U.S. jobs between 1980 and 2018. 

Workers who don’t reskill risk being left behind.

The question isn’t whether to worry about this, but how much. The Yerkes-Dodson law offers guidance: performance improves with anxiety up to a point, then declines. Just as sprinters perform best with moderate excitement before a race—neither too relaxed nor too anxious—we need balanced concern about AI.

Too little anxiety about AI replacing jobs means no motivation to adapt. Too much anxiety paralyzes strategic thinking. The right level of concern drives preparatory action. Our research suggests that regular AI use encourages this balanced perspective. Said plainly, we should neither live in denial nor flip out.

The Yerkes-Dodson law says that anxiety improves performance up to a point.

As an organizational leader, your biggest challenge is instilling that healthy amount of anxiety on the chance that AI could take their jobs — so people don’t rest on their laurels. The answer to this seems clear: exposure. We have to get people exposed to using AI more and they will naturally start to increase their interest, attention, and thus motivation to keep using.

What should you do?

Irrational Labs works with top companies to integrate AI into their organization and drive adoption and behavior change. If you’re exploring AI change management, let’s talk: [email protected].

🖱️ CLICK HERE TO DOWNLOAD THIS REPORT AS A SHAREABLE PDF.

We can help your health, finance, or tech company unlock big impact. Get in touch to learn more about our consulting services. Or learn more behavioral science by joining one of our bootcamps.