We have covered AI before.  We are its fair to say, we have been a little sceptical on the whole subject of AI.    We have sadly seen clients get quite disappointed in some of its real-life use.  This time, we are seeing AI deployed in a robot that sits on your desk.  It comes from 'Furhat Robotics' in Sweden.  For the last four years they have been building a human-like computer interface that mimics the way we speak, as well as our subtle facial expressions. The idea, according to their chief scientist; Gabriel Skantze, is that "it feels much less scary or strange compared to a more traditional robot".

We have a video below and their website is shown at the end of this post for you to check it out yourself.  There is also a reference to an english language post by the BBC who did some some useful background on this story.

The robot is being also developed with Swedish recruiting group - TNG.  The premise behind their use of it is to; reduce, or eliminate, the bias either conscious or unconscious that can happen between interviewer and candidate in a recruitment setting.  This robot specifically does not do smalltalk, although it has facial gestures and can respond with empathy.  Its designed to ask candidates the same questions in an identical manner.  Indeed even the question order is the same.  The idea behind this is that this helps reduce interview bias.

To look at the robot is simply a bald human (female) head and shoulders that can be sat on the desk.  It has a camera and can both speak and listen.  Currently it presents a transcript of the interview to the recruiter. However the aim, once it's finished development, is for the robot to make the decision on which candidates should proceed to the next stage.  

It is still in test for now (not actually being used yet for live recruitment).  But from what we have seen, and heard, the robot for a device does a very creditable job of providing a human-like interaction with the 'candidate'.  However, like other AI tools, its (AI) interactions are all based on programming and data.  It is thus only as good as the datasets and algorithms on which the programming is based.  

The two major benefits claimed are: (1) that it reduces interview bias and (2) once it is allowed to take decisions, it will be able to automate much of the time consuming interview process.   Both are noble aims.  And if you have ever faced a bulging inbox of newly applied candidates.  It may all sound very tempting.   However we are still a bit sceptical.  

First, looking at bias.  Is bias really such a big problem?  The BBC post has some good data illustrating this and on how Sweden compares in this respect by way of example with the UK.  Despite a reputation for being 'Politically Correct', Sweden has a big issue with ethnic unemployment i.e if you are minority ethnic, you will likely struggle more to find employment.  The UK by way of comparison, does not have this same issue. Or if it does its to a lot smaller extent as illustrated by the ethnic unemployment numbers.  So it does seem a lot of trouble to fix a problem that may well be, not that much of any issue for many organisations.

On the claimed automation side, we are also somewhat sceptical.  If recruitment was all about tricking a few boxes, then it would all have been fully automated years ago.  We have seen ourselves from a client AI project we have been involved with; that to get it to work at all, you need a really good - and big - dataset.  One that can be matched to the role being recruited for. This is beyond the resources of all but the biggest companies, or those who have lots of vacancies for exactly the same role.  As another qualitative consideration.  We often see client managers being swayed by the individual skills and qualities of people they come across. And swayed in a good way.  A good dialogue between manager and prospective member of staff, can help both sides work out if there is a good fit and provide confidence to proceed with a hire.  After-all its not just a one way street - good candidates are hard to find.  Candidates want to understand and meet the people they will be working with for (and ideally with).  You need to treat them right to be considered. Or this simply wont take the job.   

So, if you have been eying AI as being a tool that is ready for you. We suggest that you sit back for a moment and just think what, in practical terms, it can deliver for you today.  Then perhaps you could console yourself that you still have a job, and that - hopefully - you won't be replaced by a robot just yet...