Reading the Room · AI & the Future of Work

Computer Says No? Fifty Years of Recruitment Technology

We’ve spent fifty years making recruitment faster. Have we made it any better?

By Juli Oliver-Smith FCIPD · 4 min read

Cover for Computer Says No? Fifty Years of Recruitment Technology, with the word NO painted on asphalt

12 October 2026

If you’re of a certain age, you’ll remember how applying for a job used to work. You found the advert in the newspaper, wrote a covering letter, posted your CV and waited. Sometimes for weeks. When the reply came, it was often a no – but it came in a brown envelope, and somebody had read what you sent.

Today a vacancy can attract hundreds of applications within days. Some will have been written with AI, some will be screened by AI, and some candidates will never know whether a person looked at theirs at all.

I’ve been reading the room on how we got here, and I’ve drawn it as a timeline rather than a list, because the picture tells the story better than the words do.

What we gained

None of us wants to go back to typing letters and filing CVs in a cabinet. Each wave of technology solved a real problem. Job boards widened the reach and cut the cost of advertising. Applicant tracking systems gave us records, reporting and compliance. LinkedIn let us find people who weren’t even looking. Video interviews took the travel out of the process.

For employers, almost every stage on that timeline made recruitment faster, cheaper and easier to control.

What we moved

Look at the candidate side of the timeline, though, and a different pattern appears. A lot of the effort didn’t disappear. It moved.

We asked candidates to retype their CV into our forms. Then to write for our keywords. Then to keep a professional online presence, record themselves answering questions for an algorithm, and now to use AI just to keep up with everyone else using AI.

In the UK, the Institute of Student Employers reported 140 applications for every graduate job among its members – mostly large employers – in 2024, the highest in more than three decades. Half of those employers said candidates had used AI to help them apply.

140

applications for every graduate job

Institute of Student Employers members, UK, 2024

The loop

This is where it gets interesting. Employers automate screening to cope with the volume. Candidates use AI to produce and tailor more applications to get through the screening. More applications lead to more automation, which leads to more optimised applications.

It isn’t inevitable, but it is self-reinforcing. And it leaves both sides spending more effort to reach each other, not less.

We may be getting better at processing applications without getting any better at finding the right people.

There’s an important difference between efficiently rejecting candidates who aren’t suitable and quietly excluding capable people whose experience doesn’t fit a predefined pattern. The career changer. The returner. The person whose best skills were never in a job title.

We’ve been here before

Back in the early 2000s, when I was working at RBS, we were already outsourcing significant parts of our high-volume recruitment process – at the time we were onboarding about 300 people a month into the UK banking network. We worked with a company called Lansdowne, which managed the recruitment pipeline through application screening, telephone interviews and initial assessments, before handing shortlisted candidates back to us for final interview and selection. Lansdowne was subsequently acquired by ARINSO in 2006.

It was a sophisticated model for its time, combining technology, process efficiency and human judgement to manage recruitment at scale. And the service was excellent: it understood the organisation, its culture, needs and model, as well as working with candidates to ensure the fit.

Fast forward 25 years and, in many respects, we’re still trying to achieve the same thing: faster, more consistent and cost-effective recruitment. The difference is that increasingly we’re replacing the people who conducted those initial interviews and assessments with automated systems and AI.

But are we actually improving the quality of hiring decisions and the candidate experience, or simply reducing the cost of processing applications? And in doing so, what valuable human interaction and judgement might we be losing?

What should we be measuring?

Recruitment technology was meant to reduce administration so recruiters and hiring managers could spend more time making good decisions. So are we measuring that, or only the administration?

Time to hire and cost per hire are easy to report. Quality of hire after six to twelve months, new-starter retention and hiring manager satisfaction are harder, and they tell us far more.

And then there’s the candidate side, which most of us don’t measure at all. How long does it take to apply? How long until a meaningful response? What percentage of applicants ever hear an outcome? Did anyone get to speak to a person?

Regulation is starting to ask the same questions. The EU AI Act treats AI used to recruit and select people as high-risk, with those obligations now due to apply from 2 December 2027. That’s a recognition of something we already know: these are decisions that shape people’s lives.

Could AI give us the human bit back?

Here’s the part I find genuinely hopeful. The same technology that’s fuelling the loop could also break it – if we use the time it saves to put human judgement back where it matters, rather than simply processing more, faster.

The objective was never to process a thousand CVs in five minutes. It was to find, engage and appoint the right person, fairly and well.

So, has fifty years of recruitment technology made hiring better – or just made it easier to say no at scale? And when did you last get a reply from a human?

The 30% Rehire Problem