Both sides of hiring now use AI. Employers use it to draft adverts, summarise applications and screen video interviews. Candidates use it to write CVs and cover letters. The result is a lot of text on both sides that nobody wrote and nobody reads carefully.
Being useful about this means separating what genuinely changed from what people assume changed.
What changed for the advert
More adverts are now generated from a template and a prompt. You can usually tell: a long list of generic responsibilities, no specific description of the team, and phrasing that could apply to any employer.
The practical consequence is that the advert is weaker evidence than it used to be. A requirement list produced by a language model may not reflect what the hiring manager actually wants. This makes two things more important: finding the one or two sentences that are clearly specific to this employer, and asking direct questions at first contact instead of trusting the advert.
What changed for your application
The volume of applications per advert has risen sharply, because generating a tailored-sounding cover letter now takes seconds. Employers respond exactly as you would expect: they lean harder on the structured filters — screening questions, licences, location, salary expectations — and they skim faster.
That produces one clear conclusion. Generic AI-written applications compete with an enormous pile of other generic AI-written applications. The differentiators are now the things a model cannot produce for you: a specific number from your own work, a named piece of relevant experience, a referral, and correctly answered screening questions.
How to use AI on your side without sounding like everyone else
Used well, it saves real time:
- Structuring. Paste your messy notes about a job you did and ask for them as three achievement bullets in action-scope-result form. You supply the facts; it supplies the shape.
- Compression. Cut a rambling paragraph to two lines. Models are good at this and you are probably not.
- Interview preparation. Ask it to generate the ten hardest questions for a specific advert, then answer them out loud. This is the highest-value use by some distance.
- Comprehension. Ask it to explain unfamiliar terminology in an advert from another sector.
What to avoid: letting it invent achievements, letting it write in a register you cannot sustain in an interview, and pasting a job description in and sending whatever comes out. Interviewers notice when the person in the room does not match the writing.
Never put confidential material from your current employer into a public tool. This has ended careers, and it is entirely avoidable.
Automated interviews and assessments
One-way video interviews and AI-scored assessments are increasingly common at high-volume employers. Whatever you think of them, the practical advice is dull: answer in a clear structured way, speak plainly, look at the camera, and do not try to game a scoring system you cannot see.
Several jurisdictions now regulate automated decision-making in hiring, and you may have a right to know that a tool was used and to ask for human review. Whether it is worth invoking is your judgement, but knowing the right exists is free.
What has not changed at all
Referrals still beat everything. Specific evidence still beats fluent prose. A person who knows the hiring manager still gets read first. Turning up prepared to a conversation still wins jobs.
The tools changed. The reason people get hired did not.