The difficult evolution of automated recruitment

Since CVs became digitised, recruitment has been struggling with chronic capacity shortage when it comes to processing job applications.

 

Over successive cycles, advancing digital technologies have made applying for jobs ever-easier, leading to continuously increasing volumes of CVs and covering letters for recruiters to review.

 

The first spike in applications could be managed by early versions of applicant tracking systems (ATS), while platforms such as Indeed.com and LinkedIn made applying for a job even more intuitive.

 

Many recruiters already struggled to meet time-to-slate targets – the period from advertising a position to presenting a shortlist of two to five finalists.

 

Anecdotal evidence suggests that some recruiters had no choice but to randomly discard some job applications they received to arrive at a manageable long list – a last resort to keep afloat amid the tide of interest in open positions.

 

The boon of AI

 

Then November 2022 came with the release of ChatGPT, which, within just a couple of years, led to the rewriting of the recruitment playbook. 

 

On the upside for recruiters, generative AI-enabled applicant tracking systems (ATSs) offered enhanced candidate-matching and automated text generation capabilities.

 

The transformer architecture, the core technology behind generative AI, understands context far better than previous models.

 

Applicant tracking systems now can be entrusted with automatically scanning CVs and covering letters, checking the requirements of the job advert not only against keywords but also across their broader semantic field.

 

This deeper algorithmic understanding means that suitable candidates who fail to include the right keywords or who possess adjacent or transferrable skills would no longer fall though the cracks.

 

New ATS software also has improved capability to parse information from submitted documents, ranking candidates based on skills, competencies and personal traits.

 

Systems that harness gen AI’s text-drafting capabilities can also accelerate the writing of job descriptions and help personalise communications with candidates. 

 

Yet generative AI is not the only game changer. Predictive AI, a subset of machine learning, also has the potential to improve the myopic nature of prevailing recruitment metrics.

 

The effectiveness of filling in job vacancies is typically measured by speed: the time-to-fill metric shows how much time has passed between the job being posted and the offer being accepted by the selected applicant.

 

For a more granular view, this can be broken down further into time-to-slate (the time it takes to produce a shortlist), and time-from-slate-to-hire (which measures how long it takes to convert a shortlist into a hire).  

 

However, while drawn-out hiring processes tie up HR capacity and may risk losing a strong candidate who simply drops out, there is more to successful recruitment than velocity.

 

All the cost and effort invested in hiring a new employee will quickly evaporate if that employee quits or must be dismissed before even reaching full productivity – typically around six month for simple roles and a year for more complex ones.

 

Tying recruitment success to long-term employment outcomes can be key to extending average tenure.

 

In the UK, for example, median tenure across the service sector ranges between 2.8 and 4.4 years and employees stay for 8.7 years on average with the UK’s top employer in terms of retention rates.

 

But when employment performance data from HR information systems gets fed back into APS platforms, predictive AI can improve its selection criteria by recalibrating its scoring system against hiring outcomes.

 

Similarly to a recommendation algorithm, predictive AI could suggest candidates who are likely to perform well, fit into the company culture and have a longer than average tenure – effectively operating on a “because this profile worked, similar profiles may too” basis.

 

To keep the right balance between cultural fit and a diverse workforce, algorithms can also be calibrated to ensure that choosing similar-minded candidates doesn’t lead to uniformity in the organisation.  

 

Moreover, integrated recruitment and HR systems can be leveraged for internal recruitment too, matching open roles with employee profiles.

 

An unprecedented tsunami of applications

 

Job seekers and recruiters have always been engaged in a kind of arms race. The adage “I got the job because I exaggerated more than recruiters downplayed” captures how performative the process has been.

 

Thanks to generative AI, the number of applications submitted for a job opening have exploded in the past few years. An oft-quoted statistic suggests that in the year to June 2025, the number of applications submitted on LinkedIn had increased by more than 45 per cent.

 

This is largely thanks to generative AI’s core strength of achieving scale at the touch of a button. In this case, though, scale doesn’t increase efficiency but clogs existing systems with hundreds of applications, many of which may be irrelevant.

 

To stay on top of this chaos, recruiters have been turning to technology too, relying on more advanced ATS screening and tighter filters – which, in turn, might make quality candidates that decide not to participate in the arms race invisible.

 

It is no longer just large companies whose HR system gets overwhelmed by AI-generated CVs but SMEs too. The difference is that far fewer of them have access to ATS or HRIS tools capable of absorbing this latest cycle of disruption.

 

Although it’s hard to predict at this stage how the current misalignment between job seekers and recruiters will resolve – whether the tide will subside by itself or will trigger a shift in recruitment strategies – automation will almost certainly be a key component of the story.

 

For enterprise-sized businesses, where ATS adoption is around 70 per cent, the main challenge will be to optimise their processes through technology and data integrations.

 

For SMEs, where uptake of the technology is only at 20 per cent, the priority should be identifying SME recruitment software that meets their specific needs. What neither of them should forget, however, is the human touch, which technology is there to augment not to replace.