Anyone who’s had a hand in hiring and recruiting knows that finding the right person for the job is often difficult and expensive. Typically, organisations assess candidates using a combination of interviews and static tests, such as those that evaluate logical and verbal reasoning or behavioural traits. This process takes up significant time and resources. Moreover, studies suggest that<b> </b>without careful structuring and design, these hiring methods can suffer from low validity and fail to correctly distinguish between candidates who will ultimately be successful on the job and those who will not. The consensus is that the hiring process should be as close as possible to the experience of the actual job. This is why many organisations try to design interviewing protocols and written assessments that attempt to mirror on-the-job tasks. For example, interviews for HR specialists should include scenarios that ask candidates how they would handle conflict between employees, while interviews with marketing professionals should include discussions of marketing decisions. The problem is that asking candidates to imagine a scenario is a far cry from experiencing real workplace situations. Similarly, asking interviewers to roleplay a situation is unlikely to be realistic, as interviewers are not professional actors. All this limits how effective traditional hiring processes can be. Despite substantial investments in AI, many companies struggle to identify optimal use cases for the technology. One use case may exist in the hiring process. AI-based simulations can put candidates in realistic and dynamic workplace scenarios that require real-time, on-the-spot decisions. Such simulations can achieve the goal of creating an interview environment that is credible and close to the job environment while avoiding the substantial time investment required from human interviewers. How does this work? One example comes from Athena, a company that hires and trains executive assistants for clients who are entrepreneurs or executives. Its hiring process involves culling through thousands of applicants from several countries. On top of a battery of static tests on critical thinking, written communication and computer skills, Athena recently added AI simulations customised to the on-the-job tasks that executive assistants must perform. These simulations include a chatbot that plays the part of a demanding new client, and one that exchanges emails with the candidate who’s been assigned the goal of organising an upcoming workplace event for a client. In another simulation, candidates are presented with an inbox and must decide how to respond to emails that range from a cancelled flight to a major business crisis. These AI simulations not only administer scenarios but also evaluate the candidates’ output based on criteria such as critical thinking and attention to detail. For instance, in a preliminary sample of over 550 Athena candidates who did both the AI simulations and the static tests, 83.4 per cent of them were rejected by both tools. Disagreements between the AI tools and the static tests could either mean that the AI or the static tests are mistaken. A company could test which instrument yields better outcomes by hiring candidates who passed either instrument, and subsequently tracking their performance on the job. It should be noted that all candidates did undergo one final interview with a human to ensure compatibility, which ended up practically never changing the original decision. AI simulations can be tailored to all manner of business contexts, such as trading simulations for investor candidates, negotiation simulations for sales candidates, conflict simulations for HR candidates, and so on. Thus, at least theoretically, the applicability of AI in the hiring process is widespread. Other well-known companies are incorporating AI into their hiring journeys, including Amazon, Unilever, Delta Airlines and Siemens, among others. These organisations use AI in various ways including for analysis of video interviews and candidate resumes. In addition, new start-ups are providing AI as a service to larger firms to help them improve their hiring processes, be it by harnessing AI-powered interactive simulations to better assess a candidate’s skills or using the technology to provide personalised feedback and coaching to job seekers. There are important pros and cons to incorporating AI simulations in the hiring process. To avoid painting an overly rosy picture, we will start with the cons: Developing useful AI simulations requires precise knowledge of the types of scenarios and tasks a future employee would face in their job. This may sound obvious but, in our experience, it is not. Although most jobs have job descriptions, many leaders do not have a good enough grasp of the types of tasks of their employees perform on a regular basis. Like any hiring method, AI simulations require ongoing monitoring and testing to ensure that introducing them into the hiring process leads to comparable outcomes relative to previous methods. The technology is relatively nascent, and the vast majority of companies have not yet caught on to using AI for hiring in an optimal way, which is both an opportunity and a constraint. At this early point, it is very likely that creating AI simulations, building assessment criteria and establishing monitoring outcomes requires specialised AI capabilities that can be better provided by a third-party vendor rather than left to in-house employees. In other words, your company will have to open its hiring practices and partner with AI experts to implement this. Having said this, there are also distinct advantages, such as versatility, cost savings, realism, explainability and accuracy. AI simulations can be created and adjusted based on the needs of different companies and roles. They can reduce the necessity of humans to get involved in the hiring process until the final interview phase. This can save time and reduce the notoriously high costs of hiring. It holds the potential to mimic real-life situations that candidates would face as employees in a credible way. This can help with testing hard-to-measure soft skills. AI simulations provide written explanations for each candidate that they reject or accept, creating a trail of arguments that can aid with retrospective analysis (and possible litigation concerns, should they arise). These simulations should become more accurate over time as more and more candidates complete them. Moreover, this could potentially uncover talent that traditional assessments might miss, simply because people respond differently to static computerised tests vs. dynamic AI-based simulations, which are closer to real-life, on-the-job situations. Few would dispute the broad potential of AI. However, it is a longer and more uncertain task to identify scalable use cases in organisations. At this early stage, hiring seems to be a promising one.<b> </b> <b>This article was first published in Insead Knowledge</b> <i>Nadav Klein is an assistant professor of Organisational Behaviour at Insead</i> <i>Eliot Gattegno is the Chief Learning Officer at Athena</i>
