Identifying and managing talent in the age of artificial intelligence

7 minute read

banner 1


There are more information sources available on the subject of talent than ever before. On the digital horizon the likes of social media, AI and big data are providing organisations with a plethora of information, but does it help in the identification of and retention of talent once we’ve secured it?


Reece Akhtar, Dave Winsborough, Darko Lovric and Tomas Chamorro-Premuzic have joined forces to consider, “Identifying and managing talent in the age of artificial intelligence”.  Their subject is a chapter in the newly published book, “Workforce Readiness and the Future of Work”, a part of the SIOP(Society for Industrial and Organisational Psychology) Organizational Frontiers series.


In this blog we present a precis of the gist of this chapter and consider the implications for candidates and organisations.


It is easy to be optimistic about the future of the talent economy, but only because in some ways, it is hard for things to get worse.  More specifically, the war for talent resembles a war on talent, with organisations repelling, alienating and mismanaging the majority of their employees, including top talent. To be sure, the essence of talent is unlikely to change so long as humans inhabit the workplace. 


Akhtar et al argue that talent will always be understood in terms of four basic heuristics, namely:

  1. 1. The rule of the vital few: As illustrated by Pareto’s principle, in any group or collective of individuals, a relatively small proportion of members will account for a disproportionately large amount of group output or performance. These ‘vital few’ may be considered an organisation’s ‘top talent’.
  3. 2. The maximal performance rule: There is a well-known premise within I-O Psychology that stipulates that the best way to test a person’s ability is by evaluating the best they can do.
  5. 3. The effortless performance rule: Talented people will generally require less effort to achieve a certain level of performance than their less talented counterparts will.
  7. 4. Personality in the right place: Talent emerges from personality being in the right place. That is, when there is a strong match between people’s default predispositions and the characteristics of their respective jobs or roles, their talent emerges as a result, which fuels future career development.


Even if the jobs of the future are hard to define, it is a safe bet to expect individuals who are more rather than less able, socially skilled and driven, to excel in them.


So, how is top talent gauged? 


In the ‘Old World”: interviews, biodata collection and psychometric assessments are three popular practices:


While interviews have the advantage of the ‘human’ element to them, this can, in fact, be their main downfall: they require a significant amount of time from both the interviewer and the candidate, interviewer’s calibre can be variable, and interviewers will have their own biases, for and against, towards some candidates.



Biodata measures include information about a person’s background and life history (eg civil status, previous education and employment), ranging from objectively determined dates – date of first job, time in last job, years of higher education – to subjective preferences, such as those encompassed by personality traits.  Biodata are most commonly collected through resumes and job application forms.  The main assumption underlying the use of biodata is that “the best predictor of future performance is past performance”.  The measuring, scoring and assessment of this data are where biodata struggles.  Inconsistencies of measures, different scoring methodologies and candidate’s manipulation of content can all negatively impact on biodata’s use and validity.


Psychometric assessments

Psychometric assessments are survey-like tools that seek to measure relevant psychological constructs.  Not only do they overcome the limitations of both interviews and biodata; psychometrically developed assessments have the potential to offer superior predictive validity.  The two most commonly used psychometric assessments measure personality and cognitive ability.  Research and results suggest that psychometric and cognitive assessments, especially in combination, can generate some of the highest predictive validity rates of all selection tools.  However, psychometric assessments aren’t without their own limitations: they can be cumbersome and lengthy to complete, they can lack face validity and also rely on self-assessment and therefore raise concerns around faking-good.


Fortunately, the next generation of assessments are quickly providing solutions to these limitations.


The ‘New World’ is now beginning to see organisations using external sources to assess their talent.  The unprecedented growth of digital records is emerging as a resource for those seeking to identify or understand how best to manage talented individuals.


As William Gibson once said, the future is already here, just not evenly distributed.  This is the case with the use of innovative methodologies and tools for talent identification: some organisations are experimenting with novel approaches, but their use is patchy and inconsistent.


A recent study found that in the United States, 84% of firms use social media sites for recruitment and cybervetting is growing; 44% use candidate social media profiles to screen candidates; and 36% have disqualified candidates on the basis of information found.  To place these figures in context, social media as a predictor in hiring decisions are now as prevalent as the existing use of behavioural interviews, situational judgement tests, cognitive ability test use and growing to the use of aptitude test use.


The second shift is the transformation of these digital records and data into descriptions and predictions of behaviour, potentiating a range of tools and insights that are genuinely novel.  It is Akhtar et al’s firm belief that the effects of digitisations, although significant, are not yet profound and will grow ever more impactful.  They have coined the phrase “talent signals” to describe how individuals’ online behaviour (encompassing dynamic activity such as texts, posts, updates, tweets, photos, videos, likes, hashtags, browser use, browser history, settings, and other technical arcana; and static information such as user’s provide in profiles) represent identity claims and reliable individual differences.


Research findings are steadily accumulating and showing not just that talent signals can be identified, but used to make meaningful predictions about future behaviour.


There are three reasons the authors believe this research represents a powerful new approach to talent identification:

  1. 1. First, online behaviour is not significantly different from off-line behaviour. Research has found that social media profiles were more closely aligned with user’s actual vs. their descriptions of the “best selves,” suggesting that online presence is an extension of real-world presence.
  3. 2. Algorithmic analysis of digital data sources has the distinct advantage of a wider set of behavioural samples and consistent approaches. Research shows that machine-learning algorithms are better judges of personality than friends and family because they can process a much wider range of behavioural signals.
  5. 3. Finally, as long as organisations have robust criteria, their ability to identify novel signals will increase, even if those signals are unusual or counterintuitive. For example: understanding how the browser type an applicant uses can reflect on their initiative.  How some digital signals of personality have been used to predict purchasing activity.  How digital interviews that rely on consumer grade equipment can translate a candidate’s vocal and facial behaviours into a psychological profile or an estimate of their potential fit for a role.  And of course, there is the gamification of talent application processes.


Looking out further, the authors can imagine applications extending the range of inputs that they utilise to create more detailed insights and understanding of the people working in a company.  Blockchains might one day store all the data scattered across Facebook, LinkedIn, company databases, wearables and smartphones, allowing algorithms to interpret it and help individuals and managers determine best fit for roles, particular projects, or even team mates.  Although this may seem creepy or intrusive, if users are provided with effective controls and the ability to grant or rescind permission to access the data, the result could be a much more engaging and rewarding workplace that better fits individual talents, preferences and abilities.  Managers could be coached by their own digital assistant to change their behaviour to get the best from employees; staff could review their actions and choices to learn how to address problems and mitigate potential biases.


Getting ready for the brave new world

Talent is vital to a well-functioning economy and society – properly deployed, it drives progress for all, while offering a chance to pursue excellence and meaning to a large portion of the workforce. Increasingly accurate assessment of talent entails freedom from bias and prejudice, and capability of identifying more nuanced talent differences as well as latent talents that may blossom with the right exposure.


To realise the promise of more accurate talent assessments while minimising unintended consequences, organisations must develop four new capabilities in their HR departments:

  1. 1. Data literacy – turns data into information, and information into knowledge. The bottleneck to new more nuanced talent signals in most organisations is not the theoretical availability of the data or new solutions, but the lack of inclination and knowledge to collect, use, interpret, and trust these data to reach better talent decisions.
  3. 2. Algorithmic bias – occurs when the data used to develop and refine algorithms reflect implicit values of the society in ways that are judged as irrational or unfavourable. Detecting and dealing with algorithmic bias requires significantly more advanced data literacy, especially as new talent frameworks are being selected and calibrated.
  5. 3. Maintaining privacy – is crucial if new talent signals are going to be widely adopted and accepted. Data standards associated with new talent signals often need to go beyond existing company data standards, and a systematic approach of opting-in and developing trust will be required to ensure adoption.
  7. 4. Measuring business impact – is key to ensure that the talent signals are indeed reflecting job effectiveness.


If these four conditions are realised, the authors maintain we can look forward to a world in which talent assessments will be more accurate, and our collective talents better deployed to the benefit of us all.

If you’d like to learn more and want to purchase the book, "Workforce Readiness and the Future of Work" (SIOP Organizational Frontiers Series) 1st Edition by Fred Oswald (Editor), Tara S. Behrend (Editor), Lori Foster (Editor) click on the book cover: 

Workforce Readiness and the Future of Work book cover