A new employer survey puts a number on something HR teams have sensed for a while: the AI tools now screening most job applications are tuned for the wrong thing, and the mismatch is starting to show up in which candidates make it through.
What the survey found
City & Guilds surveyed 908 employers and 1,014 young people aged 18 to 24, including 90 who are not in education, employment or training, for its report “Counting the Cost: The UK’s Employability Crisis,” fielded in February 2026 by the research firm Deep Blue Thinking. The headline finding is not that AI screening exists. It is that employers themselves think it is filtering for the wrong signal. Sixty-nine percent of employers say AI-powered screening tools prioritize technical skills over the human skills, communication, problem solving, time management, that they say actually predict job performance. Forty-three percent already use AI to screen or filter candidates. Seventy-four percent now rate employability skills as more important than technical expertise, and 64 percent say candidates with strong employability skills have become harder to find.
That is a live contradiction inside the same hiring funnel: employers widening what they value in a hire while the automated first pass narrows on the opposite criteria.
Why the screen and the standard have drifted apart
The mechanism is not mysterious. Applicant-tracking and screening tools built on natural-language processing and keyword matching are good at parsing what is easiest to encode: certifications, tools listed, years in a title, technical test scores. Communication style, judgment under ambiguity, and the ability to manage a difficult stakeholder do not compress into a résumé keyword. So a model trained to rank candidates efficiently will, by default, over-weight what is legible to it. City & Guilds’ Andy Moss, the organization’s interim chief executive, put the mechanism plainly: “Employers are telling us loud and clear that employability skills, the ability to communicate, solve problems and manage time well, matter just as much as technical know-how, if not more. Yet at the exact moment these skills are most in demand, we’re seeing a growing human skills shortage, made worse by AI-powered recruitment tools that can filter out exactly the candidates businesses need most.”
The effect compounds on the candidate side. HR Tech Edition has previously reported that job seekers are teaching themselves AI skills faster than employers are training their own recruiting teams to evaluate them, which means the people applying are optimizing their applications for the same keyword-matching logic that employers are now discovering is too narrow. Candidates learn to write resumes for the machine. The machine rewards the resume. The employability gap the survey measures is, in part, downstream of that loop, and it is self-reinforcing: the more applicants tune their resumes to what a screening model rewards, the more that model’s rankings drift away from the qualities a hiring manager actually meant to test for when the job requisition was written.
None of this requires bad intent from a vendor or a recruiting team. It is what happens when a system optimized for throughput, screening thousands of applications against a handful of structured fields, is asked to make a judgment call it was never built to make. The 43 percent of employers who say they already use AI to filter candidates are, in most cases, using it to solve a volume problem: too many applications, not enough recruiter hours. The survey’s finding is that the tool solving the volume problem is quietly redefining the quality bar in the process, and few employers have checked whether the two match.
The generational angle employers are missing
The youth side of the survey adds a second layer employers rarely see from the hiring side of the table: 51 percent of 18-to-24-year-olds worry AI will reduce job opportunities for their age group, and only 40 percent feel completely confident in face-to-face communication, a skill nearly three-quarters of employers now rank above technical ability. Laura-Jane Rawlings, chief executive of Youth Employment UK, argued the industry is misreading the cause: “Too often, low confidence among young people who are not in education, employment or training is misunderstood as a lack of capability, when in reality it reflects a lack of opportunity, access and recognition. Addressing this requires a shift from simply defining employability skills to actively enabling them.”
Notably, apprentices in the same survey reported far higher confidence: 74 percent had received dedicated employability-skills training and 48 percent felt fully confident communicating face to face, both well above the general 18-to-24 cohort. That is evidence the gap is trainable, not innate, which changes what “fixing the AI filter” should actually mean for a hiring function: less about tuning the algorithm and more about what gets built into the pipeline before a candidate ever reaches it.
What it means for the HR leader
Three implications follow directly from the data rather than from speculation about AI in general. First, an AI screen tuned purely for technical-keyword match will systematically reject candidates an employer says it wants, which is a validity problem, not just a fairness one; a screening tool that filters against your own stated hiring criteria is a broken tool by the employer’s own definition. Second, this is not a hiring problem alone. HR Tech Edition has covered how the skills-first hiring era is running into its own credibility test, and this survey is another data point in that same pattern: stated criteria and screening mechanics keep diverging. Third, the apprentice comparison suggests the fix is upstream of the applicant-tracking system, in what candidates are taught and how employers describe the role, not downstream in filter settings.
The practical response is to audit what the screening tool is actually optimizing for against what the job description and hiring managers say they need, then build structured, human-scored steps back into the process at the exact points, initial screen, technical interview, cultural interview, where the AI-only pass is doing the most damage. Skipping that audit and trusting the vendor’s default ranking is how a company ends up with the exact mismatch this survey documents: a stated hiring priority that the hiring process itself works against.
Source: City & Guilds
