A job description is supposed to explain what a company needs. Increasingly, it also explains who the company imagines doing it. That second message can be much quieter than the first. Nobody has to write an age requirement. Nobody has to say that an older applicant would be unwelcome. A handful of familiar phrases can do some of the sorting before a résumé ever arrives.
A new national survey from CWI Works, conducted by Ipsos, puts numbers around that problem. Among 1,842 U.S. adults ages 18 to 75, 76% said people assume older workers are less comfortable with new technology. Among adults 55 and older, that figure rose to 81%. Yet the survey found a very different picture among older workers who had actually used artificial intelligence on the job: 87% said they were at least somewhat comfortable using it, and 49% said they were very or extremely comfortable.
That gap between stereotype and behavior is interesting on its own. The more revealing part is what happens when the stereotype gets translated into copy.
CWI Works asked how workers 55 and older react to language commonly found in job descriptions. A phrase such as 'recent college graduate' is obvious enough; 69% said it would make them less likely to apply. But the softer language matters too. 'Digital-first' or 'digital native' made 41% less likely to apply, and 44% read the phrase as a signal that the role was intended for a different age group. 'Culture fit' carried that message for 46%. 'Go-getter' did it for 45%. 'High potential' did it for 36%.
None of those phrases has to contain the word age to create an age-shaped picture. That is what makes them useful as a case study in social sorting language. The employer may believe it is describing energy, adaptability, technical fluency or growth potential. The applicant may hear something else entirely: we already have a person in mind, and that person is younger than you.
'Digital native' is the cleanest example because it disguises an identity assumption as a skill requirement. Being born during a particular technological era is not a competency. Knowing how to use a particular system is. Understanding AI tools is. Learning a new platform quickly is. Evaluating whether an automated output is trustworthy is. Those things can be demonstrated, taught and measured. Nativeness cannot. The phrase takes a capability an employer could describe precisely and replaces it with a generational shorthand.
That shortcut becomes stranger as the workforce itself stops cooperating with the stereotype. The survey found that 71% of Gen X respondents had used generative AI, essentially level with the 74% reported among Millennials. Workers 55 and older who were unemployed and looking for work were also among the groups most interested in opportunities to update their skills. The people being quietly told that the future may not be for them are, in many cases, already using the future and asking for more training in it.
This is where the problem becomes structural rather than merely impolite. Employers routinely describe talent shortages, skills gaps and difficulty finding qualified candidates. Then the first public-facing artifact in the hiring process can narrow the pool with language that has little to do with the actual work. A company can spend heavily on recruiting technology, sourcing platforms and employer branding while losing candidates because somebody wrote 'digital native' when they meant 'comfortable learning new technology.'
The same issue sits inside 'culture fit.' Companies generally do need people who can work effectively with colleagues, operate within shared norms and contribute to a functioning team. But 'fit' is elastic. It can describe collaboration, or it can become a vague test of familiarity: does this person look, sound and behave like the people we already picture succeeding here? The vaguer the language, the more room there is for candidates to interpret the invitation through whatever signals the company has accidentally supplied.
That matters because applying for a job is already an exercise in self-selection. Recruiters do not only reject applicants; applicants reject themselves. People decide whether the salary is worth it, whether the commute is possible, whether they have enough of the listed skills and whether the organization seems like a place where they could plausibly belong. A few words in a posting can change that calculation long before an employer has the opportunity to assess the person behind it.
The fix is not to drain every job description of personality until it reads like an appliance manual. It is to make the language do more actual work. If a role requires someone who can use AI, say what they will use it for. If the company values learning speed, describe the changing systems the person will be expected to learn. If initiative matters, explain where the job requires independent judgment. If teamwork matters, say what collaboration looks like. Specific language gives candidates information. Coded shorthand asks them to infer whether they are the sort of person the company had in mind.
There is a broader lesson here for workplaces that goes beyond age. Organizations often think of bias as something that happens at the decision point: who gets interviewed, promoted, trained or hired. But systems make decisions earlier than that. A form can make one life easier to describe than another. A policy can assume one kind of household. A job ad can quietly describe an identity while pretending to describe a skill. By the time a human decision-maker enters the process, some people may already have received the message.
CWI Works' findings are especially useful because they catch that message in an ordinary place. There is no dramatic prohibition to point to, just familiar workplace language doing more than its author may realize. The employer thinks it wrote a description of the opportunity. The reader is also looking for a description of who belongs inside it.
Sometimes bias does not look like a rule. It looks like copywriting. And if companies genuinely want the broadest pool of capable people, the first question may be simpler than another recruiting platform or another AI hiring tool: what, exactly, did we tell them when we asked them to apply?
SOURCE NOTES
Survey figures attributed to CWI Works / Ipsos materials cited in SOURCE NOTES. Cultural framing is RMN's.