Has AI made job applications too easy? Should recruiters put some friction back into hiring?

‍AI has made applying for jobs dramatically easier, and hiring teams are paying the price in application volume. The answer may be to deliberately make parts of recruitment harder. The useful kind of friction asks candidates to demonstrate knowledge, identity or ability before you invest heavily in assessing them.

For most of the past decade, recruitment technology has pursued one fairly consistent objective: remove friction.

Reduce the number of clicks. Shorten application forms. Let candidates upload a CV rather than re-enter their employment history. Make jobs easier to discover. Make applying possible from a phone.

There were good reasons for doing this. Long, repetitive application processes lost good candidates.

AI has changed the calculation. A candidate can now find vacancies, tailor a CV, draft a covering letter and answer application questions in minutes. Some tools can automate significant parts of the process entirely.

In July 2026, The Irish Times reported on a business analyst vacancy that received more than 2,000 applications. The recruiter quoted said a similar position might previously have attracted 100 to 150. Applications were also arriving within 24 to 48 hours.

That creates a strange problem.

Organisations spent years making it easier to apply. We may have succeeded too well.

More applications do not necessarily mean more choice

Receiving 2,000 applications sounds like an abundance of talent. For a recruiter, it can mean the opposite.

You still need to identify the handful of people genuinely capable of doing the job. Now you have to find them inside a much larger pool of submissions, many of which have been optimised using the same tools.

AI can turn a weak application into a polished one. It can incorporate language from the job description. It can produce competent answers to predictable screening questions.

Has the CV become worthless? The CV and covering letter therefore tell you less than they once did. That does not mean somebody using AI is dishonest. Using software to improve an application is hardly new.

The problem is signal. When almost anybody can produce a strong-looking application quickly, the application itself becomes less useful as evidence of suitability.

Your hiring process has to find the signal somewhere else.

Some friction is useful

Recruiters understandably dislike the word friction. You associate it with seven-page application forms, duplicate data entry and asking candidates to create an account before they can even submit a CV. Nobody needs those things back.

Useful friction has a different purpose. It creates a small cost for applying while generating information that helps you decide whether someone should progress.

Imagine you are recruiting for a specialist compliance role. You could accept a CV and an AI-written covering letter from 1,000 applicants.

Or you could ask one specific question:

“How would you respond if a senior employee asked you to delay escalating a potential sanctions match until a commercial deal had closed?”

A strong candidate should have something meaningful to say. The question gives you a better signal than another paragraph about being a “results-driven compliance professional”.

The same principle applies to technical roles. Ask someone to solve a short problem related to the actual work.

For a customer-facing position, ask for a concise response to a realistic customer situation.

You are adding friction. You are also giving serious candidates an opportunity to distinguish themselves.

Identity may need to move up earlier in the process

There is another question employers now need to ask much sooner: Do you know who you are actually interviewing?

Traditionally, identity checks have tended to sit near the end of recruitment. You choose a preferred candidate, make an offer and then complete background screening. That sequencing made sense when the main purpose of identity verification was to support pre-employment checks.

AI creates a different problem.

If applications, written answers and even parts of interview preparation can be generated or heavily assisted by software, employers have less certainty that the person progressing through the process is the person represented by the application.

Remote recruitment adds another layer. You may never meet a candidate physically before an offer is made.

That means identity verification may need to serve two purposes: supporting background screening and establishing that you are assessing the right person in the first place. For some roles, waiting until an offer is made may be too late. If you are about to put a candidate through a second or third interview, involve senior employees or ask them to complete a meaningful assessment, there is a reasonable argument for verifying identity first.

That does not mean asking every applicant who clicks “apply” to complete an identity check. The timing should be proportionate.

A more sensible point might be after initial screening and before a second interview, particularly for remote roles, senior appointments, regulated positions or jobs involving access to sensitive systems and information.

You can then proceed with greater confidence that the person you are spending time assessing is the person they claim to be.

We've written before about how to trust who you are hiring when AI can manufacture convincing candidate information. An identity check can increasingly be part of the selection process itself, rather than something reserved exclusively for post-offer screening.

That is a meaningful change.

Friction should increase as certainty becomes more valuable

You do not need to put every candidate through a two-hour assessment. That would replace one bad hiring process with another.

Friction should increase as someone gets closer to the job.

At the beginning, one or two well-designed knockout questions may be enough. They should test something genuinely relevant rather than whether the applicant has discovered the correct keywords. Later, a short skills exercise can establish whether the capability described on the CV exists in practice.

Before you invest in further interviews, an identity check may establish that you are dealing with the person you think you are.

A structured live interview then lets you probe the candidate's reasoning rather than scoring a perfectly constructed written response.

Credentials deserve the same treatment when they matter to the role. A qualification written on a CV remains a claim until somebody verifies it with the relevant source. For regulated and specialist positions, professional qualification verification can be far more valuable than another round of subjective interviewing.

The principle is simple: ask for more evidence as the consequences of getting the decision wrong increase.

The danger is making candidates perform unpaid work

There is an obvious objection to strategic friction. Employers have a habit of asking candidates to invest ridiculous amounts of time before deciding whether they are genuinely interested in them.

A three-hour presentation for a first-round interview is not clever screening. Neither is asking a marketing candidate to produce a usable campaign or a designer to complete half a client project. Good friction should be proportionate.

If you ask 500 applicants to complete a 90-minute assessment because your recruitment team does not want to review CVs, you have transferred your application-volume problem to candidates.

That damages the experience for exactly the people you want to attract.

The same standard should apply to verification.

An identity check introduced at the right stage can reduce uncertainty. Asking every early-stage applicant to complete one regardless of role or risk would simply create another barrier.

Ask yourself what additional information each step gives you. If you cannot explain how it changes a hiring decision, remove it.

Recruiters may have to stop optimising for application numbers

There is another uncomfortable implication.

For years, a large applicant pool has looked like success. Recruitment dashboards show application numbers. Employer branding teams celebrate reach. Job platforms tell you how many people have clicked “apply”.

AI makes those numbers increasingly meaningless.

Ten well-matched candidates who have demonstrated genuine interest, ability and identity can be considerably more valuable than 1,000 generated applications. The metric that matters is the quality of the evidence you collect about each serious candidate. This requires a change in thinking about candidate experience too.

A good candidate experience does not mean making everything effortless.

Candidates also benefit from a process that distinguishes genuine expertise from mass applications. Strong candidates should want an employer to assess the things they are actually good at and to run a process where everyone is being assessed on the same terms.

The frustration comes when the work feels pointless.

Hiring may become slightly harder on purpose

There is an arms race developing in recruitment. Candidates use AI to generate and submit applications at scale. Employers respond with AI to screen them at scale. Candidates then use AI to improve their chances of passing those systems. Keep following that logic and hiring becomes two sets of software exchanging increasingly polished information about a human neither side has actually evaluated.

The more automated that exchange becomes, the more valuable verified human signals are likely to become.

Ask candidates to prove a few things that matter. Ask a question that requires experience. Watch them solve a relevant problem. Verify who they are before you invest heavily in interviewing them. Speak to them live. Confirm the credentials on which your hiring decision depends.

Every one of those steps adds friction.

That may be exactly why they are becoming useful again.