What does the future of talent assessment look like given AI is making inroads?
The three key criteria for successful talent assessment are: attractiveness for candidates, efficiency of the entire process and validity of the hiring decisions. In all three areas AI can generate significant advantages. The attractiveness of a selection process to candidates is mainly driven by being maximum short, transparent and engaging. AI helps making the selection process shorter by taking more information into account and by including passive data of candidates. AI increases the efficiency of the entire process by automation. By using natural language as main communication between candidates and the recruitment system, candidate’s engagement can be ensured over the entire process. Ai-based natural language processing is already used but will increasingly dominate the measurement process and the selection workflow.
Humans or bots and AI - who can hire the best candidate?
Simple answer: both have advantages but an optimised cooperation between humans and AI will beat any of the two alone. So far, AI is typically not used to automate jobs, rather to automate tasks and augment human functions which in turn increases productivity and performance. It is unlikely that at any time recruiters will become superfluous through AI. Recruiters will become more efficient, more productive and more successful by using AI. However, the recruiter's job will change and become more complex and responsible because recruits will control highly complex, AI-driven processes. Our future is about collaboration between humans and machines.
See video: the changing role of Recruiters in an AI world: https://youtu.be/eldGZNkItm8
AI seems to be the holy grail of talent assessment? What are the pitfalls? What should organisations be careful about?
The thing people typically forget about artificial intelligence is that it’s exactly that — artificial. Behind the most powerful algorithms are vast, complicated datasets, which are built and labelled by a vast human workforce. These tasks in HR can be enormous and there is a tendency in organisations to believe that it just needs enough data. But the point is that the size of the available data doesn’t matter if the quality of the criterion data is flawed, because it is biased and hasn’t a conceptual framework that takes into account future requirements. If there is one tip: start with proper job analysis that considers future requirements and avoid to simply map on criterion data.
What are the ethical and legal challenges of AI?
If AI is an essential part of a psychometric instrument, the same challenges and quality criteria are applied. Therefore, objectivity, reliability and validity build the legal foundation for making any process defensible that involves AI. And as we have seen for unsupervised Online Assessment, enhanced quality criteria become relevant: fake-resistance, hardware-independence and acceptance.
With AI, the quality criteria are enhanced again by controllability and transparency. According to GDPR (the European General Data Protection Regulation) we have to be very careful if data is captured as a by-product of a measurement process that is not relevant for the diagnostic process, but can be used to flag candidates, e.g. if an AI detects that someone wears glasses in a video interview and infers a medical attribute. This is the reason why we consequently follow a ‘glass box’ approach when deploying AI. This means that the entire process is construct-based, fully transparent and therefore legally defensible.
How is AI changing the way companies appraise their employees?
Performance appraisal is an area where AI-based people analytics in combination with passive and unstructured data, can generate new levels of insight into KPIs of an organisation. Already more relevant data on an employee exists in an organisation that is not yet linked to performance. AI-based people analytics allows the overall metrics of the organisation to be tracked through to the contribution of individual employees. The appraisals themselves will be on multiple measurement points, multi-view references, combined with passive data and then directly connected to follow-ups, next steps, potential analysis, development paths, and trainings. AI-based machine learning will link diagnostics and trainings closer together for everyone’s benefits.
Learn more about AI in Assessment: https://assessment.aon.com/ai-in-assessment/
About the AuthorFollow on Linkedin Visit Website More Content by Achim Preuss