At this time of year, there is a flurry of activity in the field of recruitment and hiring. Artificial intelligence, automation and machine learning have made their way into the all aspects of the employee journey, including recruitment and human resource management. This raises a host of questions. Where and when can AI be used in the employee lifecycle? Will automation eliminate or entrench bias in the decision making processes of HR? And will AI reduce the numbers of HR professionals we will need in the future?
Recruitment, is an obvious entry point for AI into HR as it the first step in the employee-employer journey, and one that is very resource intensive. Several start-ups have moved into this space in the last few years, and for some reason, many have human names such as CharlieHR, EVA.ai and JAMIEAi.
According to Wilson Wong, head of insights and futures at the Chartered Institute of Professional Development, algorithms are currently being used to sift job applications and to quickly create candidate shortlists. He explained that AI is sold by companies that create automated tools as a way to increase productivity and efficiency. The corollary of is that organisations can reduce the amount of ‘touch’ in the recruitment process, “because human touch is expensive.”
Wong made it clear that these algorithms in this context do not constitute smart machines as they do not learn, but rather look for and pinpoint desirable patterns. Charlie Markham, CTO and COO of EVA.ai also made a clear distinction between artificial intelligence, and machine learning and chatbots. The former, he sees as simply a marketing term while the latter two are the technologies that his company make use of.
He said that machine learning allows recruiters to make really good decisions at scales that would otherwise be unmanageable. “Machine learning enables people to find needles in haystacks using large, multidimensional datasets and to use historical datasets,” said Markham.
He said that his clients tell him that the main benefits of automating the recruitment process are not just reductions in costs but also reductions in the time to hire, whereby a three-month cycle could be go down to two months, one month or even one week. “That’s much better than any saving on agency fees or HR internal people costs,” he said. Markham also explained that a recruiter could increase the number of candidates they look after with the help of EVA’s tools. “Your typical recruiter can manage 1,000 to 5,000 candidates at a time. With EVA you can easily multiply that by 10 to 20. I’ve seen one recruiter manage a pool of 50,000 candidates, and this was a relatively junior person so you can probably go higher.”
Markham said EVA has a bot that would free recruiters from making registration calls, doing screenings and booking interviews, thus giving them more time to approve the candidate to be booked for interview. “Rather than doing the legwork themselves, they approve stage transitions. They become much more approvers and supervisors rather than doers so it becomes a much more high level job for them,” Markham said.
To Adrian Ezra of Jamie.AI, the positives of applying AI to hiring are clear and almost identical to those identified by Markham. He said that by using an automation tool in recruitment, companies can eliminate the costs of recruitment fees, use fewer man-hours, scale faster, reduce the time spent by the company and avoid the productivity cost of spending a lot of time making a hire.
Wong added that learning and development is another area of HR to which automation can make a significant positive impact, especially if coupled with virtual reality. He said: “If you are setting up a learning environment, automation allows you to dip into all kinds of resources, to mix and match, to explore interactively with other learners outside the organisation. That kind of potential is huge.”
He gave the example of a company that was able to reduce training time for apprentices from six months to two or three weeks. Instead of shadowing a senior worker for half a year, the apprentices would interact with tools and machines in a simulation cockpit through virtual reality while still meeting with senior technicians daily. “The productivity gain of that kind of learning is incredible,” said Wong.
In spite of these benefits, Markham called for caution when applying machine learning. “The application should be made in a supervised capacity, not unsupervised. This is where mistakes have been made.” He cited the case of Amazon which had to scrap an AI recruitment tool in 2018 after it ranked male candidates higher than female candidates. The tool did so because it had been trained on the CVs submitted to the tech giant over a 10-year period, reflecting the discernible and problematic gender imbalance in the tech world. This serves as a lesson to those training algorithms, to not assume that what good looked like in the past, is necessarily what it should look like in the future.
Markham said that supervised machine learning requires constant monitoring and checking of what goes into and what comes out of the “bucket,” and ensuring that quality and fairness are embedded.
In relation to bias, Wong’s cautionary tale warning against reliance on humans to do the right thing when teaching algorithms is that of Microsoft’s Tay. In 2016, the Twitter chatbot mimicked the deliberately offensive language of some Twitter users and as a result spouted racist and sexist responses.
He said: “While we shouldn’t say that artificial intelligence is a solution to everything because its programming is still programmed by flawed human beings with their assumptions and unconscious biases, neither should we say the solution is human because humans beings murder, lie, are unfaithful and are unpredictable.”
One point on which Wong and Markham do not agree is that of the need for HR professionals to be data literate in order to use the automation tools that could help them in their jobs. According to Markham: “Machine learning shouldn’t be discernible from magic for the end user. It is a mistake to say that everyone needs to learn about machine learning because that is entirely unrealistic.”
Instead he said that the onus on the companies deploying AI in HR to ensure that it works correctly and is unbiased. He said: “The diligent, careful providers like us guarantee the quality and fairness of our algorithms. Then it will get to the point where everyone has machine learning, some realising it and some not, but all enjoying the benefits.”
This is the polar opposite view of Wong. He said: “There are concerns about data literacy in the profession and hopefully curricula in HR will keep up to date with technology, because technology does require you to look under the bonnet.” He also said: “AI will figure quite heavily in the future because intangible value is where businesses make their money. They make their money off data, analysing that data then using that data to predict buying or behaviour. If HR is going to remain relevant, practitioners really need to understand how these algorithms work to ‘provide insight’ on predictive behaviour.”
However, when automation and machine learning is rolled out on a massive scale, will there be any HR professionals left or will the technologies lead to massive redundancies? The experts were unequivocal. According to Ezra, the ultimate aim is to replace them. “That’s kind of the whole point,” he said. “As the industry moves towards more automation, the result will be fewer agencies, fewer recruiters but that doesn’t mean you won’t have very good recruiters doing what they do very well.” Both Markham and Wong share Ezra’s view that it is likely that automation will lead to fewer but better HR professionals.
In Markham’s view the problems associated with traditional HR, like bureaucracy, slowness and getting the incorrect candidate, are eliminated when solutions such as EVA are implemented. “HR professionals can become more strategic, more like advisors, rather than having to deal with the nitty gritty. It can make HR faster moving and more dynamic,” he said.
Wong took a wider view when looking at the future of HR. He said: “The governance role of HR will grow. The ethics side of it will grow. The question is whether HR can transform itself so that it is more data-savvy and more ethics-literate in terms of assessing these decisions. My own gut feel is the profession will be smaller but actually more relevant if it can make that transition.”
Source link
The post The impact of artificial intelligence on human resource management appeared first on HireGDPR.
