On the meaning, sense and use of AI
Back in 1990, Ray Kurzweil wrote about the ‘age of intelligent machines’. At that time, Artificial Intelligence was like a futuristic sci-fi concept. Today, AI is everywhere, proving Kurzweil was right. But let us not be blinded by what AI is, does and will be able to do. Only when machines are analyzing, learning and taking autonomous decisions, we can speak of AI. Having said this, AI has an enormous potential in recruitment technology. What’s more, Actonomy is at the forefront in implementing it. Read about the what, why and how in this white paper by Filip De Geijter, CEO at Actonomy.
Ray Kurzweil was a visionary, yet lightly arrogant computer scientist. In his must-read ‘The Age of Intelligent Machines’, he explains the Law of Moore, as initially formulated by Gordon Moore, one of the founding fathers of Intel: the speed of computer chips will grow exponentially, whilst the size of them will shrink at the same speed. What’s more – and what made him a real futurologist – is that he predicted what the result would be: the upcoming merger of computers and humans.
So far, his predictions have proven to be correct. Artificial Intelligence, as one of the corner stones of his theories and views, has come from nowhere to almost everywhere in almost no time. Smart chips are even entering our bodies (like in the case of some diabetes treatments), Artificial Intelligence is making its entrance in more and more applications.
Let’s start by defining Artificial Intelligence
So there we are, Artificial Intelligence slowly entering our machines, our devices, our bodies, our lives. Internet of Things, Big Data, Robotics: they are taking their part in the technological evolution. The field of human resources is not an island, and therefore AI is taking a slice of the cake there as well. But before turning into listing some wow-experiences and ‘you won’t believe this’ phrases, let us first try and be precise in the terminology. That is: what is Artificial Intelligence?
Apart from the objective statement that it is hard to define what ‘intelligence’ is (and, by the way, who is intelligent), there seems to be a relative consensus on the definition of AI. Take the famous ‘Turing Test’ as a starter: if a computer is capable of fooling people and making them believe the computer is a human, than this computer must be intelligent. We from our side follow the vision of German professors Andreas Kaplan and Michael Hänlein:
‘AI is the power of a system to correctly interpret external data, learn from them and use them to realize specific goals and tasks.’
At Actonomy, we rephrase it by summing three prerequisites that need to be fulfilled before we can speak of AI: a piece of software that can analyse data (data analytics), learn from it and make predictions (machine learning) and finally take decisions.
Analyze, predict, decide
Beware with this definition: the sting is in the tail. Often, systems developed in HR technology are capable of analyzing huge amounts of data. The area of big data is not so new anymore, and many computers are strong and fast enough to handle large datasets, often from different origins, of incremental nature and in diverse formats.
So far, so good. Learning and making predictions is one step further, as we see in the use of so-called expert systems. Here, systems use statistics to predict what the most probable next step is supposed to be in a given context, without the use of explicit knowledge. This is the domain of concepts such as neural networks. Third, algorithms must be capable of taking decisions, in order te be called artificially intelligent.
Our ‘HR-ontology’ is our big data
Now, let’s turn back to our own playground: the world of semantic searching. At Actonomy, our technology is based on a large amount of data (our ontology in the specific domain of job titles and skills), where our algorithm is capable of analyzing links and make predictions. For example: if it encounters the word ‘software engineer’ and ‘object-oriented programming’ in a given CV, it might conclude that this software engineer knows Java.
Third, our software is deciding what vacancies are best suited for the candidate who in this example is a software engineer. Analyzing, predicting and deciding, all at once. At the kernel of this piece of AI-based searching and matching technology, lies our incrementally growing ontology, by far the largest and most elaborated HR-ontology in the industry today.
The difference with other semantic search engines
Our algorithm is not a black box where you just get decisions as output. Rather, ours shows the whole trajectory followed in order to come to a certain decision. That is, you can easily backtrace why our matching software matches a candidate to a job posting. You can understand the reasoning behind the match. This transparency is far from evident.
A few weeks ago, we tested a commercially available job matching engine using semantic searching from a different technology provider. After having entered our CV, we did not get any single relevant job offer. Neither did we get any insight why the specific (unsuitable) offers were made: there was no backtracing. A missed opportunity, if you ask us. Artificial Intelligence should be transparant to be trustworthy.
If we utilize the definition of AI as set out before, we must be hard when judging where and how AI is really used in current HR-technologies. People in HR tend to believe the two extremes of the scepsis spectrum. On the one hand, AI is said to solve all problems. Let us be clear: AI will not. On the other hand, they fear AI will replace human jobs. Let us be clear: AI will not. Stripping all extremes and pub talk off from reality, Artificial Intelligence sometimes looks more like ‘Artificial Hype’. But to be really meaningful, technologies in general, and AI in specific, must pass from ‘nice to have’ to ‘need to have’.
Nice to have or need to have?
It is important to realize what we can achieve when implementing AI in HR-processes. And also what we cannot. Take a recruitment robot. If applied smartly, such a robot will certainly be helpful in doing the cumbersome, tedious parts of screening interviews, whether it be via a human-looking robot or a straightforward chatbot. But in order to make smart decisions in a recruitment process, the human factor will remain crucial. That is why we are sure AI will – at least in the short term – not replace human recruiters.
Rather, thanks to AI the recruitment process will become more human: human recruiters get more time available to do face-to-face interviews, check motivational aspects, discuss expectations, etc. People in the recruitment industry must not be afraid of AI. The invention of the hammer has not replaced the job of carpenter. Rather, thanks to the introduction of the hammer – this wonderful tool that is a pure ‘need to have’ – the carpenter has specialized and become more of a craftsman adding his personal touch to his woodwork. Innovations forces people to become more creative and add human value.
Only by honestly understanding what AI is, we can examine the usefulness of it in the HR-world. A correct understanding of the concept AI will prevent it from being just another technology hype. Take the recruitment robot: should it have the form of an actual talking face, or is it enough to just have a chatbot? What is needed, what is nice?
So, what will be next in for AI in HR?
Let us conclude by going back to Ray Kurzweil. His predictions on Moore’s Law only went till 2011. It would be extremely interesting to extrapolate his ideas and see where we get when it comes to the further integration of AI in our daily lives. Self-driving cars seem to be a given fact. But what will be next in, say, human resources? Will the recruiter in the end become obsolete, because recruitment robots will not only screen on hard skills but also take over the soft skills and cultural behavior? Luckily enough we know for sure: Moore’s Law will slow down because of the fundamental physical barriers when it comes to chip sizes. But let this not make us fall asleep.
Because whilst being critical-by-nature and skeptical by character when we talk about AI in HR, we must remain vigilent. As a technology developing company, we also must be well prepared. Just like we did 15 years ago, when we started using semantic ontologies in our technology, we will remain at the forefront of what the future will bring.
Even if we cannot see it (yet). Just to leave you a bit uncomfortable, one – wild? – idea about this AI-future: will artificial emotion – AE – be the next step? Algorithms capable of analyzing, predicting and deciding emotions: will they ultimately take the last domain we still believe to be the unique domain of humans in recruitment? We do not know. But at least, we think about it, and prepare. Preparing: just another pure human capacity that makes us different from smart computers.