Long reads

AI vs HR

25 Jul 2017 By Georgi Gyton and Robert Jeffery

Chatbots can already perform basic HR roles, and what’s coming next is a game-changer. But that doesn’t mean your job is under threat – as long as humans ask the right questions before it’s too late

There are almost certainly more than five million Amazon Echo devices now humming away happily in homes across the globe, with some estimates of the product’s soaring sales to date putting the figure at nearer nine million. This sleek, voice-activated tower might not do much more than connect your home devices, open apps and access the internet, but it looks and feels like the future.

At the heart of the Echo sits Alexa, a personal AI assistant who acts as the device’s brain and voice. Leaving aside the fact that Alexa is more a product of smart marketing than actual artificial intelligence, people love her (she’s received, recorded and declined 250,000 marriage proposals in her short life).

It is just four years since the film Her, based on the premise of a relationship between a human and an AI home assistant, was positioned as a piece of science fiction. Yet now, a more platonic union seems far from fantastical. Indeed, for Robert Bolton, partner in KPMG’s Global HR Centre of Excellence, the real question is not why anyone would wake up with an Alexa, but why you’d leave such technology at home when you went to the office.

Soon, you won’t have to. Bolton points out that it’s already possible to integrate Alexa with an HCM system and have a conversation with her about HR matters. He gives the example of a woman on maternity leave who tells Alexa about her new arrival, when she’ll return to work and when she wants her baby to be added to her insurance records. The update is instantaneous: no HR professional has been involved, and the employee hasn’t had to access any software herself.

AI and HR are having a ‘moment’ right now. “It’s as if a lot of HR leaders got an Echo for Christmas,” says Bolton. “Last year, they weren’t really engaging with the topic, while finance and customer-facing parts of the organisation were. Now, HR has woken up.”

Peter Cheese, CIPD chief executive, agrees. AI-enabled automation, he believes, is more significant than previous waves of automation, and has the potential to “reshape” HR. “This is one of HR’s biggest agendas right now,” he says. It is both a threat – since in HR, as elsewhere, it has the potential to mechanise huge numbers of tasks and entire roles – and an opportunity to take ownership of a smarter way of working.

But what exactly is the opportunity? AI is a spectrum term that has come to take in advanced data analytics at its most basic (essentially, the ability to handle a question by looking up an answer in a data set) through to an emerging category of machine learning, defined as the ability of an algorithm to spot a pattern in data sets and use it to intuitively predict what will happen next. But true AI – software that answers its own questions before a human has asked them – remains possible only in theory and may not happen for two generations or more, if ever.

What exists already is the processing power to scale up AI rapidly. “The technological ability to compute at the scale and speed needed for widespread AI has now been achieved,” says Steve Wells, futures analyst and COO of AI research business Fast Future. “At the same time, the rise of cloud-based digital services via Amazon and Google have essentially outsourced AI and put the technology in reach of even the smallest organisations.”

Fast Future estimates that while less than 5 per cent of jobs could be fully automated by adapting currently available technology, 35 per cent of the UK workforce does a job that is at risk of potential, albeit partial, AI automation.

Around £1bn was invested in UK-based AI firms in the first half of 2016. PwC thinks AI could add £232bn to the country’s GDP by 2030, the equivalent of a 10 per cent boost. It is particularly significant for the UK, since it aims squarely at automating or altering tasks performed by white collar workers, which is why journalists, insurance underwriters and accountants are among those worried it is coming for their jobs. Who needs a junior lawyer, for instance, when a computer holds the entire history of case law and has enough ‘intelligence’ to work out which parts are relevant to your case?

The canary in the coal mine of such technology is customer service. Lloyds and RBS both offer chatbots – text-based applications that carry out a ‘natural language’ conversation by accessing a database of predetermined phrases – as the first point of customer contact. American Express has integrated one into Facebook Messenger for accessing accounts on the move. A start-up is building a chatbot it thinks could replace the NHS’s non-emergency 111 number.

And now, the UK is being graced by its first HR-focused chatbots. Software provider MHR says it developed its bot for clients after noticing how popular internal messaging platforms had become. It wanted a ‘playful personality’ that would appeal to millennial employees.

“Very quickly, answers to frequently asked questions – such as ‘when is the next pay day?’ and ‘how many holidays do I have left?’ – can be built into the chatbot, which means HR can get on with more important strategic tasks,” says Tim Johnson, MHR’s chief commercial officer. The bot works by accessing the existing HR self-service system rather than deploying machine learning. It will soon be able to sync with a calendar or handle an expenses claim, a not-dissimilar service to one offered by Indian firm Acuvate, which says it has an ‘HR assistant’ chatbot ready to launch in the UK.

Johnson says the bot ‘saves’ the equivalent of one HR professional per 1,500 employees. But it won’t be long before such technology is making a grab for more than just the prosaic, self-service aspects of HR. Take CoachBot, an application built by UK start-up Saberr and trialled by organisations including Unilever, the NHS and NFU Mutual. It aims to revolutionise the sort of team-building work traditionally done by an external trainer, by asking team members simple questions about their communication style, work goals and team dynamics.

Having built profiles of each employee, it begins to make suggestions about how to work smarter or handle conflict, linking to L&D content that might help. While Saberr admits a chatbot can’t help address fundamentally broken working relationships, it is confident that letting staff interact individually, at a time that suits them, is an innovative solution.

To see where this might take us, you need to head to Islington, less than a mile from London’s Silicon Roundabout. Above an RBS branch, inside a cavernous former office block being refitted as a co-working space, the team of data scientists at Satalia are far busier than the preponderance of novelty mugs and office toys might suggest.

Satalia uses machine learning to help organisations with scheduling, predicting customer behaviour and advanced logistics. It works for the likes of University College London and Odeon, specialising in ‘complex problems’. And when it comes to HR, Satalia wants to walk the walk. Given that the company is “data informed, not data driven”, says head of human operations Jai Clarke-Binns, it wants to use its technical expertise to make the way it manages people fairer, easier and more efficient.

“AI is everywhere without you knowing it,” says Clarke-Binns. “It’s infiltrated our lives.” HR, like the rest of the business, will have to “pivot massively” to keep up with such technology: “You can’t say the skills you need in HR today will be the same tomorrow at all.”

The good news is that it will be the ‘boring’ tasks the technology comes for first. AI, says Clarke-Binns, could be used to create reward packages that are individually relevant, or to act as a coaching assistant for new managers: “Imagine if you could use AI to coach the individual through the journey, to make sure they are asking the right questions, monitor the work they are doing or onboard new starters, for example.”

Satalia is already internally beta testing a programme that pits its finest minds against the terror of the annual pay review. Why, it wondered, was so much power over reward decisions handed to individual managers? Couldn’t AI make a fairer, more objective decision?

The end point of this musing is a system whereby staff input how much of a pay rise they feel they should receive – all salaries are already transparent – and other employees use the system to vote for whether all, some or none of it should be accepted. Satalia has a ‘network map’ showing how frequently people interact with each colleague via internal collaborative tool Slack and coding system GitHub – it weights each vote according to the closeness of individuals’ working relationships, and their seniority and centrality, controlling for other factors that might distort the figures or any attempts to ‘game’ the process. The result, it hopes, is an entirely bias-free pay round.

It is a laudable dream. But Satalia is also keen to look at other areas of HR, such as the potential for more intelligent recruitment. Already, there are commercial applications on the market that sift CVs and use algorithms to supplement or subvert human decision-making. If you’ve applied for a job in the past year, there’s a fair chance AI has handled your CV.

There are understandable concerns about the propogation of bias in such processes (see right) but the entire basis of machine learning is that errors can be rectified over time. Indeed, the next wave of AI will hand more power over HR decisions to computers. It will be able to pick up on common problems: if employees are confused by a new annual leave policy and are failing to complete holiday requests, it may be able to suggest that HR improve its communications on the issue. Or it might mine comments on internal social networks to create what has been called an ‘emotional dashboard’ of stresses and strains, or pockets of good practice.

“A chatbot is just a medium of communication,” says Johnson. “But [eventually] we could use predictive analytics to highlight to leaders of an organisation what’s changing in the dynamics, or be able to see where the bad leaders are in a very transparent way.”

This hands huge power to AI – perhaps too much. “The ability to make decisions for us, particularly ones that can be made based on our past history, is one of the key promises of AI,” says Wells. You might give your personal AI the chance to choose your benefits based on what it knows about you, he says, but would you trust it to negotiate your salary with your boss’s AI, even if it promised to do a better job than you could? If, as some believe, AI could become your ‘digital twin’, would you let it sit in on meetings and represent your views?

For Bolton, part of settling such questions is rethinking the structure and the remit of the HR department. HR business partners, in his view, might become the ‘front end’ of HR, backed by a vast amount of AI making the transactional work happen. Or it could be AI ‘chatting’ to staff on the frontline while a centralised bank of humans acts on the intelligence.

The billion-dollar question is whether such technology will ultimately enhance or displace HR professionals. Will it mean more empowered and meaningful work for employees? Or will HR follow the customer service route, where the number of roles is already being eaten away at: with an Accenture study claiming 80 per cent of customer queries can be handled by a chatbot, the future of the call centre looks bleak. If you hope to move HR up the ‘value chain’, can you possibly need anywhere near as many people when you get there?

This is a philosophical debate being played out on a grand scale – and the truth is nobody can be sure. Will we follow the trend played out during previous industrial revolutions, where new categories of jobs emerge to broadly compensate for those displaced by new technology, a notion championed by MIT professors Erik Brynjolfsson and Andrew McAfee? Or will the view of author and futurologist Martin Ford, in Rise of the Machines, prevail? “You can imagine lots of new industries,” he has said of the decades ahead. “But they won’t employ many people.”

The UK’s Sutton Trust has predicted that AI will be disastrous for social mobility at work and may “disrupt” up to 15 million jobs. “Nobody knows exactly what will happen, but it seems complacent to suggest that everything will be the same as it has been,” says Cheese. “I genuinely think this is different. If we look at it in a positive way, if we use technology well and are smart in its adoption, it will allow us to work smarter, not harder, remove the less appealing parts of jobs and focus on adding value. But to get there, you have to consciously design the use of technology – you can’t assume it’s going to happen.”

Even tech evangelists are hedging their bets that some sort of universal basic income – which guarantees a subsistence level of income for those in advanced economies – might be necessary if enough jobs are eroded. In HR terms too, there is an acceptance that the function’s dual focus on strategy and empathy is not easily replicable but that it cannot possibly be untroubled by such profound processing capabilities.

This may not be a bad thing, says Bolton. “I’m passionate about HR being much more evidence-based. I’m passionate about the strategic value of people management, but I’m also highly critical of what’s been done in the name of HR that has no evidence whatsoever.

“This technology can be a lot more evidence-based, and can create a more frictionless way of engaging with the enterprise. It can encourage the appropriate balance of help from machines and help from humans… and it can finally allow HR to perhaps fulfil that strategic role it has always sought.

“The question is who owns the data. If HR retains ownership of people data, it continues to have a role. If it loses that, all bets are off.”

Those in the function who are purely focused on process should be worried, he adds. They should “learn to be more relevant” to survive: “Think about your skills and competences and what you can bring to a digital world. That doesn’t necessarily mean going out and learning coding – though that might be the answer for some – but you need to develop STEM skills and get creative in the more traditional sense of the world.”

HR’s other role in an AI future will be to help make decisions about if and when to automate, whether to reskill or redeploy the human workforce, and the moral and ethical aspects of such decisions. Bolton describes an insurance company that until recently was “like the wild west”, with people experimenting with bots with no thought for the implications. Now, it has realised that HR should be central to the governance of AI automation.

Not every business is as enlightened. “A lot of the debate around technology and automation in the workplace is being driven by things like productivity,” says Cheese. “The fact that the price point of AI and robotics is coming down so much is very important. Technology that was seen even two or three years ago as extremely sophisticated has now become affordable.

“HR has to understand the agenda much more, and be a greater part of the debate. How do we get the best out of our technology and our people? If we believe in a better future for technology in the workplace, we have to design for that.”

Even so, there are reasons not to be so fearful of the future. In their new book, Brynjolfsson and McAfee point out that even start-ups assume the structure and flavour of traditional firms, because it remains the most efficient way of doing business.

Cheese, too, is optimistic. “People can work alongside technology,” he says. “It can enhance what they do, enable them to do more meaningful work and use their inherently human skills. Increasingly, the skills we will need are critical thinking, empathy, collaboration and creativity. That’s what we need to keep promoting, because, not only are they hard to replicate, they make us who we are.” And besides, who else will AI employees turn to when they have a grievance with their android line managers?

Join the debate about the technology of the future on The Future of Work is Human platform

HR Adviser

HR Adviser

Holborn and Covent Garden, London (Greater)

£35,850 to £40,975 per annum + excellent benefits

The National Housing Federation

HR Manager

HR Manager

Canterbury, Kent

Competitive Basic + Bonus

Automotive Company

HR Business Partner

HR Business Partner

Leeds, West Yorkshire

Salary circa £39,900 per annum

Leeds Trinity University

View More Jobs

Explore related articles