CEO and Co-founder, Dr Alex Linley, looks ahead at 2020 as he shares three key themes this will shape the landscape for talent assessment.
Looking out over 2020, I share my thoughts on three key themes that will shape the landscape for talent assessment in the coming year.
#1 – The Data Subject Strikes Back
Privacy has been a concern that has been a long time in the making for modern society, but it is now clearly with us. The European Union led the way on this with the introduction of the General Data Protection Regulation on May 25, 2018 (I remember it only too well…) and all the data subject rights it enshrined, but that privacy movement has now very much made its way across the Atlantic as well.
The California Consumer Privacy Act took effect from 1 January this year and upholds similar rights for data subjects as GDPR in Europe. Given that this legislation comes from the state that is the home of Silicon Valley, the irony of the new law is not lost on many people.
If privacy and data subject rights are now being recognized as legitimate concerns in Silicon Valley’s home state, then the world is indeed changing. Change is afoot, since the data subjects are indeed striking back, and the business model of surveillance capitalism may need to evolve as a result.
#2 – All’s Fair in Love and…Recruitment
Fairness has always been a core issue for recruitment and selection, but the status quo of how fairness is defined or measured is being challenged like never before. US law enshrined the four-fifths rule which for years has been the bedrock of standard assessment practice, but now is no longer enough.
Indian economist and philosopher Amartya Sen distinguished between procedural justice and distributive justice, or fairness of process and fairness of outcome, respectively. While we can argue that each of these approaches are ‘fair’ in and of themselves, they can lead to quite different results in practice. This is at the heart of the fairness debate – do we want a fair process and a potentially unfair outcome, or a potentially unfair process but a fair outcome. The law as it stands, and society as it is becoming, are often in disagreement on this.
Applied to recruitment and selection, this leaves us faced with the questions of whether we ensure (for example) a 50:50 balance of male and female candidates being offered (which is fair in terms of distribution, but may be procedurally unfair); or whether we follow an appropriate legally equal process, but accept that the distribution of outcomes (e.g. the proportion of male and female candidates hired) may not be in perfect balance.
These questions are further complicated by yet more different conceptions of fairness, especially within machine learning and artificial intelligence (AI). These include individual fairness (based on the predicted score for an individual) and group fairness (based on the distribution of scores within two groups), as but two examples. When one understands that these different definitions of fairness can be inherently in conflict with each other, it’s easy to see that fairness in recruitment and selection will continue to be a key theme for years to come.
#3 – Data, Data, Everywhere, and Not a Drop to Drink
It started with the Hollerith punch card, progressing to the humble database and before too long we had ‘Big Data’ and ‘data warehouses’. Now we’re talking about ‘data lakes’, which for many organizations could equally be ‘data ponds’ or even ‘data puddles’, given just how fragmented and unconnected our data sets often are. Whereas once we had so-called relational databases (that didn’t actually show the relationships between variables), now we’re increasingly using non-relational databases (which paradoxically are often defined by their ability to map the relationships between variables).
This new-found interest in data from organizational clients is often predicated on their desire to use AI, with the recognition then following that AI needs data in order to operate and build models, and so the first thing we need to do is to ensure we get our data in order.
The second thing that often follows is then working out what the data outcomes are that we are trying to target. This is where companies often fall down when they realize that their current performance evaluation metrics simply won’t stand up to any sort of meaningful scrutiny.
For 2020 and beyond, we expect to see more and more focus on cleaning up data, collecting the right data, and the design and implementation of systems for collecting the right data outcomes in the right way that serve overall organizational goals.
#4 – Bonus! – I Can See Clearly Now – As 2020 marks the beginning of a new decade, and particularly with the bad jokes we can make about ‘2020 vision’, we will see lots of predictions in 2020 for the coming years. As with all predictions, astrological or otherwise, we should treat them with a healthy dose of skepticism.
