by Alistair Cox@Hays

By 2020, it’s predicted that businesses across the world will spend a combined $47bn on artificial intelligence (AI), and it appears that every aspect of our lives – from shopping and leisure to work and personal finance – will be transformed as machines leverage data to provide us with tailored, personalised services at scale.
根据预测,到2020年,全球企业将在人工智能(AI)上投入470亿美元,似乎我们生活的方方面面 — 从购物、休闲到工作和个人理财,都将随着机器利用数据为我们大规模提供量身定制的个性化服务而经历变革。

With such a wide-ranging impact, it’s only natural to expect that AI will also revolutionise the way that organisations search for and acquire talent. Automated technology can analyse the mountains of data across organisations and the wider job market, translating it into easily digestible formats which will ultimately help humans make better decisions and spend their time on higher value, higher impact tasks.

So what effect will this have on the way we source talent in the future? I see at least three areas of evolution:

One of AI’s main benefits is that it allows processes to be completed at a rate and scale that is simply unachievable by humans. Therefore, I expect two particular subsets of AI to become widely adopted over the next few years:

  •  Natural Language Processing (NLP) – transforming text into structured, easily digestible data – it effectively lets a computer read language
    自然语言处理(NLP)— 将文本转为结构化且易于理解的数据 — 有效实现计算机解读语言
  • Natural Language Generation (NLG) – the reverse of NLP, transforming structured data into text – letting a computer write language
    自然语言生成(NLG)— 自然语言处理的反向过程,将结构化数据转为文本 — 实现计算机生成语言

Both NLP and NLG have enormous potential in talent acquisition. The digital age has brought huge benefit to our industry. But it has also brought massive quantities of data that currently is handled largely manually. A simple job ad for example can elicit tens of thousands of responses, many of which may be wholly inappropriate applications, yet all must be screened in order to find the real stars. Straightforward yet often time-consuming tasks such as CV screening, drafting job descriptions and communication with candidates could take a matter of seconds using this new technology. That should free up the human experts to spend more time on the valuable role of working with the very best candidates on a personal basis – in effect putting the relationship back into the role, as that’s the crucial element required for success.
NLP与NLG都在人才招聘领域中展示出巨大潜力。数字时代让我们的行业受益匪浅。但随之而来的是现今大部分仍需人工处理的海量数据。例如,一个普通的工作招聘广告可能会收到数以万计的回应,其中许多申请人可能完全不适合该职位,但为了找到出色人选,所有的申请都要挨个筛查。筛选简历、起草职位描述以及与候选人沟通这类简单却耗时的工作,利用AI技术只需几秒钟就能完成。这样一来,人力专家就能腾出更多时间与最优秀的候选人进行单独交流 — 这实际上让招聘变得更加人性化,而这正是成功所需的关键因素。

Here at Hays, we’re already utilising NLG AI in the candidate screening process with an external expert organisation in this area, and the early signs are that it works. This platform has certainly accelerated the shortlisting process for us, and it also enables our recruitment consultants to concentrate on assessing the individual candidates outlined by the technology to be the best fit for the role in hand, rather than pouring over a wider pool of hundreds of thousands who may not be suitable. Our consultants are freed up to concentrate on building relationships with their clients and candidates – something that definitely can’t be done via AI.

There’s a lot of emphasis today on eliminating bias in the recruitment process. I welcome that, but it is scientifically proven that unconscious bias can still exist even when great efforts are made to eliminate the more visible routes to bias. It’s interesting to see how, as well as increasing efficiency, automating sections of the screening phase can also lead to a decrease in subconscious hiring bias. After all, if the AI system is instructed to compile a shortlist by focusing solely on data around a candidate’s role suitability, it will by definition ignore demographic information such as age, race, and sex. There are pitfalls to beware of though, and it would be wrong to assume AI is inherently fairer. It’s equally possible for machine learning to automate an existing bias, and the “black box” nature of many AI techniques could mean you’re unaware of this even happening. For example, if the historical data used to train an AI screening algorithm had an inherent age bias in it, removing age data alone from the input files might not fix the problem as there are cases where AI can infer age based on first names and their popularity over time. As with any new technology, there’s a lot more to it once you start to get involved and develop it and beware the unintended consequences. Never forget that AI is only as good as the data you feed it, and to compile your ideal shortlist of candidates, you’ll need to provide comprehensive criteria on which qualifications, previous experience and specific skills would appeal to you.

AI could also enable businesses to emphasise candidate fit like never before, which should ultimately result in more successful, long-lasting hires. After all, we see that the number one cause for an unsatisfactory hire is a lack of cultural fit between employee and organisation.

This is already happening to a certain extent with skills – online job boards increasingly use algorithms to match their community of candidates to available roles. For example, a LinkedIn job posting will rank candidates by matching the information listed on their profiles to those in the job description. However, as AI (and the data collected by businesses) becomes more sophisticated, we can expect to see these algorithms become more complex and take preferences and fit into account, not just technical capability to fill a role.
招聘市场上使用AI计算匹配度已不是新鲜事 — 在线求职平台越来越多地使用算法来进行候选者群体与开放职位之间的匹配。比如,领英职位招聘会通过匹配候选人主页信息与职位描述信息的方式来给他们排序。然而,随着AI(以及企业收集的数据)愈加精益复杂,这些算法也将朝着更复杂的方向发展,算法会将招聘偏好与匹配度考虑在内,而不仅仅只看技术上的能力。

Individual’s attitudes to benefits, company culture and salary preferences amongst other aspects can be assessed through survey metrics. Machines can scour the jobs market and process answers via an algorithm to provide businesses with a shortlist of candidates that match their organisation’s persona. Candidates themselves will also feel the effects, and you can expect vastly more accurate job recommendations and a more tailored outreach from prospective employers, honed even more closely to your preferences.

However, the human element will still be required, probably even more so, since it remains incredibly difficult for any machine to analyse the soft skills that remain so crucial to modern business. I’m yet to see an algorithm that can read things like humour, temperament or enthusiasm as effectively as a person can. And let’s not forget that ultimately human oversight is still required to compile criteria – I certainly wouldn’t want a machine deciding the persona of my business, and I don’t think it would do a particularly good job yet.
然而,人类仍然是不可或缺的因素,且重要性可能还甚于从前,因为任何机器都难以分析对现代企业至关重要的软技能。我至今还没发现可以像人们一样有效地理解幽默、气质或热情的算法。请切记,筛选标准最终仍需人类的监督 — 我绝不想要机器来决定我企业的人格,我也不认为现阶段机器在这方面做得足够好。

Aside from helping businesses hire the right candidates today, I believe AI will play a significant role in enabling organisations to retain and develop talent for the future.

We’ve seen the retail sector harness AI to prompt and nudge consumers with more personalised and interactive shopping experiences. I expect that in the coming years employers will follow suit, keeping staff engaged on a more specific, one-to-one basis. Again, the overall use for AI here is to supplement, not supplant, human management – an automated system could prompt a manager to catch up one-to-one with an employee who values frequent mini-reviews, or remind them that there is one member of staff who hasn’t yet been included in an internal reward programme. These are very basic use cases. As algorithms get more sophisticated, employers may, for example, find the machines telling them when and where they are likely to lose valuable people so that humans can intervene before it’s too late.
我们已经看到, 零售行业正利用AI创造更具个性化和互动性的购物体验,促使消费者购物。我预计未来几年,众多雇主也会追随其步伐,在更具体化、个人化的基础上保持员工的参与度。再次重申,AI在此领域的整体应用是为了补充、而不是替代人类的管理 — 自动化系统可能会促进经理与看重小型回顾会议的员工进行一对一交流,或提醒经理哪位员工尚未得到内部奖励项目。以上都是十分基础的用例。随着算法变得越来越复杂,机器可能会告诉雇主何时何处可能会出现人才流失等信息,以便人类在为时已晚之前进行干预。
One exciting prospect is to utilise AI to supplement proactive human planning. An organisation’s talent flow is essentially another data spread that a computer can analyse to spot upcoming trends, either assessing when future revenue growth will require additional staff, or analysing calendar patterns to identify which time of year employees are most likely to depart, for example. I believe that this will become integral for larger businesses, who will then work with hiring managers and talent acquisition leaders to plan proactive hiring initiatives, rather than spending so much time on reactive, ‘fire fighting’ hiring.

As AI’s business case becomes more widely recognised, I expect that talent acquisition will begin to adapt very quickly indeed. Not only should this result in a far more efficient recruitment process, it will also provide talent managers with more time to focus on higher value tasks – and opportunities to leverage ever-important human nuance, which I believe can never be successfully emulated by AI. However, expect a deluge of new businesses and models claiming that they can overnight transform your fortunes as that is usually the bandwagon that starts rolling once a new idea starts to gain traction. It will not be easy to sift the real jewels from the rest of the noise and recruiters could well find themselves overwhelmed with choice and uncertainty on which route to take. Certainly in my own business we spend a lot of time looking for the ideas that will really make a difference for our clients and candidates, and that’s probably less than 10% of all the proposals that come across our desk.

However I also firmly believe that people, not machines, will continue to play the dominant role in hiring and staff engagement. We will need to set the criteria, we’ll need to bring that magic of human nuance to the screening and interview phase, and we need to build the person-to-person relationship which is all that ultimately matters when a candidate holds their pen over a contract. Today people do business with people and I hope that never changes. Despite the excitement and fears around the rise of AI, talent management largely remains a contact sport, where gut feeling, grounded in thousands of tiny facets of human experience which are never captured as data, plays just as strong a role as hard data.