ArticleRecruitment industry

Why AI won’t kill the recruitment industry

After co-founding a business like TempaGoGo, which brings tech to help traditional recruitment agencies, it’s always instructive to take five minutes out now and then to see which predictions of the industry’s imminent demise at the hand of some new technology have been disproved.

Take for instance Oleg Vishnepolsky, the Global CTO at DailyMail Online and metro.co.uk, who wrote an opinion piece back in 2016 that started “I reckon the industry has about 2 years left in it, 4 at most”.

Hmmm – it’s still here. And growing.

When looking at the state of recruitment-industry-killer-technologies, it’s important to not confuse the excitement around an announcement of investment in a start-up, or hiring of AI experts, or trial of some new algorithm, with the actual success, or otherwise, some time later in the field.

I spent nearly 7 years as the IT Director of Reed Specialist Recruitment – one of the UK’s largest chains of traditional agencies. It’s a family-run business, with its founder Sir Alec Reed, now 85, still very much involved. One of the things that he teaches is that “only fools forecast”. If he had listened to the continuous predictions of the next recruitment-agency-killer technology, perhaps Sir Alex would have shut up shop years ago. Rather, he embraced the tech to Reed’s advantage. History shows us forecasting is hard to get right.

Three False Sunsets

Image of a sunset
Many people have forecast that new technologies were going to wipe out the recruitment industry – but history has proved them wrong

The job boards

Take for instance the rise of the job boards, allowing candidates to submit their CVs and apply for jobs, posted directly by employers for a little over £100, to avoid those agencies with their wretched high fees.

Job boards flourished: now, there are many employers who do-it-themselves, using these platforms to reduce costs by cutting out the agencies.

But those very agencies are also some of the job boards’ biggest clients. It turns out that job boards are a super efficient way of finding candidates for agencies as well.

Furthermore, some companies have found the experience of direct hiring very expensive in terms of their own precious time: sorting through the applications, reading lots of CVs, responding to applicants that weren’t short-listed, organising interviews, convincing the favoured candidate to join, managing the whole process – all this is very time consuming, especially when you might receive one to two hundred applicants for some roles; or, on the other hand, receiving none at all and wondering what to do next. Agencies are good at handling all those kinds of activities and situations – hence why companies still use them.

Social media platforms

Next came the social media platforms – or shall we just say Facebook. Some predicted a world where companies would be able to find the perfect candidates and connect with them through the platform, targeting them thanks to the information gleaned from their profiles. It didn’t quite turn out like that. 

Worse in fact. Some candidates found that sharing too much of their personal lives so publicly didn’t enhance their employability, as some potential employers sifted through their profiles and found off-putting posts. It can be hard to reconcile ‘that’ image of a person as the ‘light-and-life’ of a party, with the ideal candidate you want to fill the accounts payable role you have open.

Facebook, to this point in time at least, have used all those user profiles to build a super-targeted advertising platform, and did not manage to disrupt the recruitment industry. That might change but at present, they probably have their attention diverted to adapting to a world that is a little more concerned about giving too much data away than it used to be and trying to police unwanted propaganda / fake-news on their platform. Coming up with new disruptive recruiting products maybe isn’t high on their priority list.

Nonetheless, recruitment agencies have used Facebook to extend their online presence and attract candidates. Just as with the job boards, the agencies have been able to use a technology, that some felt could put a dent in their bottom line, to their advantage.

LinkedIn

Around the same time LinkedIn arrived. It took more time to build its presence, but around 5 years ago it was actively being talked about as the next mortal threat to the recruitment industry. Just as with job boards and Facebook, employers where going to use it to find candidates by themselves, and cut out those pesky agencies once and for-all.

But – and you probably know where this is going – it turns out that agencies are now some of LinkedIn’s biggest clients. The same issues that came up as using a job board apply here too. It takes knowledge and time to use the platform effectively – something that recruiters now need as part of their skills, but not something every manager with an open vacancy has to time to learn and perfect on top of their normal duties.

Therefore, one could draw the conclusion that all of the tech that was predicted to diminish, if not destroy that recruitment industry, has actually strengthened many agencies, and helped them find individuals that would have been out of reach from their existing talent radars. 

So the recruitment industry lives to fight another day.

The Next Contender Enters The Ring

Image of boxing gloves
Will AI be able to do what other technologies have failed to deliver and reduce our reliance on traditional recruiters?

But hold on a minute, what’s that technology on the horizon. Or if you read the tech press – a bit closer than the horizon. Machine Learning – Artificial Technology – AI… this looks more promising?

Surely, this is the one that’s going to see the ‘Office Space to Rent’ signs go up on agency buildings. It’s going to solve the issues job boards and social networks didn’t address. AI is going to not only find candidates, but find the best, and maybe handle the interactions too.

Going back to the article by Oleg Vishnepolsky – he referred to the news that LinkedIn had acquired a startup called Connectifier, and had employed a number of machine learning experts. AI is being developed to try greatly to reduce the utility of recruitment agencies, by finding the best candidates and proactively present them to clients.

Will it work? Well, I’m going to heed Sir Alec’s advice and not forecast… It’s definitely easier to be an historian! But I will venture that I think there are three things the tech will need to do well – none of them trivial by the way – to make a credible impact on the recruitment industry. I believe such tools will enhance recruitment agencies, not diminish.

I also see opportunities in the ‘lower hanging fruit’ basket, that technology can help with today whilst AI improves and helps ‘tomorrow’ – but more of that later.

It’s Tough Being Human

A large crowd of people
Deep human attributes, such as judgement, empathy and learn experience are very hard to replicate with AI

There are three tasks the humans are particularly well suited to, that machines need to work on: judging individuals, selling and handling people-based issues.

Judging individuals

The first of these we could narrow in scope: judging candidates against a client requirement. Much of today’s research has focused on trawling through a huge database of CVs to find candidates that would be suited to a role that is described by a pure text document. It has some success, but to take root, there are two really hard bits still to contend with.

Hiring managers might not have time to write down, or find it hard to articulate, all they really need from their ideal candidate. They might have downloaded a template role description from the web, added a few unique skills that they require and top it off with a mention of their corporate’s culture. It’s tough to write down precisely what you need – what’s in your head. And whilst algorithms might do a fair job of matching, even looking at resumés of other candidates that hold such roles to similar people – the results are just not sufficiently strong to stand alone and require a human to catch errors. And then, there are those it may have missed – perhaps candidates that didn’t know how to make their resumé stand out to the algorithm, but would have been the best person for the job.

The other element to consider comes with bias: if you present the same CV for the same role to two different people to judge, they may come to different conclusions. One for example might have worked in large organisations and value people that come from a similar background, when on the other hand, the second interviewer thinks that people from small companies have a better can-do attitude. It may be hard to admit, but when we read a resumé, our judgement is directed by a huge number of invisible factors our experience has woven in. Getting AI up to the level where we trust it is as good or better than us is going to be tough. Do we want to it to come to the same decision we would have, or can we trust it when it differs?

Sales

The second hard skill for AI to learn is sales. Because a large element of recruiting is all about sales: selling roles to candidates, and selling candidates to clients. I have yet to come across AI that is good at selling to people. Let’s not think Amazon’s “People that bought this, also bought this”, or automatic chat bots on websites are going to cut it.

Being a good sales person is to understand about what each party values, read implicit signs, (such as gesture, tone of voice, or choice of words) and then using ethical techniques to persuade such that over the long term both client and candidate are happy with the outcome. It’s a non-trivial tech task for a computer to record such information, process and use it in a natural manner and be at a level of confidence that would stand up in court, ??should delete word?? say a candidate allege they have been discriminated against by the algorithm.

Handling people-based issues

Finally, if we want AI to take over from a recruiter, it’s going to have to get good at dealing with humans and all their faults. In the world of temp recruitment, dealing with situations where the candidate doesn’t turn up, measure up once on the job, or is treated badly by a client, are all part of the job, requiring good judgement, empathy and tact. They’ll need to handle all of this – if we truly think of the complete set of services the recruitment consultant provides.

All this illustrates why I think a non-trivial set of problems need to be solved to really kill-off the recruitment industry as we know it. It’s easy to over-look all the tasks a recruitment consultant actually does. No doubt elements of this technology are going to help make recruitment agencies more effective and efficient – just as the job boards and social platforms have.

Yet let’s assume tech can do these thing – and yes I believe it will get there – do we really want it?

Getting A Leash On It

Dogs pulling at the leash
We need to be thoughtful as what AI we are going to be happy to deploy and hand over some our more precious tasks

Google’s recent impressive demonstration showing an automated call handling service, that could book a restaurant table for you or handle an incoming call, show that such tech is very much in-bound. See this video for a demonstration of what it can do: https://youtu.be/-qCanuYrR0g

Immediately after its release, a healthy debate was triggered about the degree that we are ceding control to tech, and whether this is a good thing. 

One might argue that having too much AI involved in hiring is going to introduce unintended biases, frustrated clients, suspicious candidates. Companies at their most basic are groups of people acting to fulfil some mission. Who do you want to choose these people?

In response to concerns over AI, Google has recently tried to establish an independent council with the catchy title: Advanced Technology External Advisory Council (ATEAC). It was going to be dedicated to “the responsible development of AI” – but has proved so hard to gain acceptance of who should sit on the board it has gone back to the drawing board (no pun intended!).

I think that this all shows that just as the technology is going to be hard to develop, so is the law and acceptance by the general population.

Back To School

A range of coloured pencils
Whilst we wait for AI to earn its place as a recruiters tool, there are plenty of opportunities to improve the industry using well known tech today

So whilst we wait for AI to solve these problems, debate their use and utility, there are actually still lots of opportunity to improve recruitment for clients, candidates and agencies using so ‘old-school-tech’.

Although the UK has the second largest temp recruitment sector in the world, it still relies on some pretty basic solutions. Take for instance timesheets.

Every week a timesheet will need to be submitted by a candidate, signed by a client and then processed for billing and payroll to take place. But do you know what – some recruitment agencies will ask for spreadsheets to be emailed, or worse faxed back. OMG.

Given the hugely fragmented nature of the market, there’s no easy way to request a temp, review candidates online, or interact with agencies. Other markets have solved these problems – think price comparison sites for holidays and insurance, or websites that aggregate houses on sale.

So rather devote all our energy into developing AI to automate some tasks which are super hard for computers, there are literally dozens of opportunities we know how to solve today that shouldn’t be starved of resource. We can use the proceeds and knowledge these generate to invest wisely into AI that can work happily alongside recruitment consultants, not replace them.

And that’s what TempaGoGo is all about. 

It’s going to be fascinating watching the industry adapt, as it has done over the last two decades – and I look forward to having this front row seat.

Leave a Reply

Your e-mail address will not be published. Required fields are marked *