AI has changed hiring on both sides simultaneously, which has produced a strange equilibrium: applications are faster to write and harder to differentiate, while screening is faster and no more accurate.
What employers actually use it for
CV parsing and ranking. Applicant tracking systems have extracted and matched keywords for years; the newer generation summarises and scores candidates against a specification. Note the important nuance: most systems rank and surface rather than auto-reject, so the goal is being read favourably rather than defeating a filter.
Screening conversations. Automated video and chat interviews at first stage, more common in high-volume recruitment.
Sourcing. Searching candidate databases and professional networks for matching profiles - which is why a well-optimised LinkedIn profile has become more valuable than it was.
Assessment scoring, particularly on structured tasks.
Interview scheduling and note-taking, which is genuinely useful and uncontroversial.
What candidates use it for
Writing CVs and cover letters, preparing for interviews, researching companies, and in some cases generating large volumes of applications automatically.
That last use is the problem. It has raised application volumes sharply while lowering their average quality, which makes employers filter harder - which pushes candidates to generate more applications. Everyone works harder for the same outcomes.
The practical consequence: generic AI text no longer differentiates
When most applications are competently written by the same handful of models, competent writing stops being a signal. What now stands out is exactly what AI cannot produce: specific, verifiable detail about your actual work. Numbers, named projects, decisions you made and what you learned.
Use AI to structure and tighten your writing. Do not use it to generate content you did not supply, because the resulting text reads as generic to an experienced recruiter - and increasingly to their tooling.
How to adapt
- Use the posting's exact vocabulary, because both software and humans match on it
- Front-load specific, quantified achievements - these are what survive both automated and human filtering
- Keep formatting plain so parsers read your CV correctly: no columns, tables, graphics or text in headers
- Optimise your LinkedIn profile for the terms recruiters search, since AI sourcing runs on it
- Prepare for automated first-round interviews - structured answers, clear audio, specific examples
- Expect more live and in-person assessment, added specifically to verify what written applications now cannot
The fairness questions worth knowing about
AI screening trained on historical hiring data can reproduce the patterns in that data, including discriminatory ones. This is a live regulatory issue - the EU AI Act treats employment-related AI as high risk, and UK employers remain bound by the Equality Act regardless of what tool made the decision. Candidates in some jurisdictions have rights regarding automated decision-making. If you believe you were rejected by a fully automated process, it is reasonable to ask.
For employers, the statistical point is the important one: a model trained on past decisions learns past biases, and measuring your own process is the only way to detect that.
What has not changed
Referrals still work best. Specific evidence still beats claims. The ability to explain your work clearly to a person is, if anything, more valuable now that written applications are a weaker signal.
The durable position
Skills that combine technical capability with human judgement are the ones least exposed to this shift. The Statistics for Business course at London School of Business UK builds analytical judgement rather than tool operation - see also our AI courses for leaders. Enquire today.