In April I wrote an article about how Artificial Intelligence (AI) is playing an increasingly central role in recruitment, transforming how organisations find, evaluate and hire talent. It highlighted how some of us are being slow on the uptake and that we need to get moving.
Well, our candidate market is ahead of us. As executive recruiters, we are receiving an increasing number of AI-written CVs, and experiencing more and more ChatGPT-assisted interview responses, resulting in significantly more of our time being spent exploring and verifying those truly appropriately skilled and experienced, and weeding out those who are not.
Therefore, we have to continually review and enhance our own recruitment processes to overcome the challenges we face with the use of AI. Here I have some top tips for you in order to do the same.
Whether you use an application form or accept CVs, a key is to incorporate structure into what you require from prospective candidates. Ask for specific achievements or metrics, for example: ‘Describe a time you improved a process'. Include measurable outcomes. Incorporate limited length answers to reduce the effectiveness of copy-paste AI-generated content. And as I highlighted in April, look out for overly-polished or generic language, repetitive structure and formatting, American spelling and date formats, lack of personalisation, etc.
There are AI-detection tools in the market where you can run applications through AI-generated text detectors which flag overly generic or suspect content, but we know these to be very much still in trial phases and not fail safe. Funny that, being AI.
Bespoke assessment tools where there are time limits can be customised and role-specific of course. Timed logic tests, situational judgement tests and ‘real-world' problem-solving scenarios make it harder for candidates to rely on AI in real time. However, even then, in an age where we are conducting assessments remotely and/or online, there is room for using external tools such as AI.
The response to that is to introduce browser lockdown to the testing interface; this effectively prevents candidates from accessing other websites, platforms, or functions such as copying and pasting. Alternatively, consider webcam monitoring where you can capture live video footage of the candidate and their immediate surroundings while undertaking their assessment exercises.
When it come to the interview process, we at Osborne Thomas do not embrace automation, all interviews are conducted live – whether face-to-face or virtually. We ask situational or behavioural questions that require complete spontaneity, and we probe further with follow up questions based on previous responses to comprehensively test any candidate's depth of understanding; the approach to follow up questions being along the lines of ‘why, how, explain in more detail' etc. If you are an organisation in favour of embracing automation and therefore likely to be using asynchronous video interviewing as a tool, be sure to give candidates only a few seconds to respond and limit/prevent any retakes.
With final interviews, we suggest more of a focus on the softer skillsets, their cultural fit, their values and behaviours and their ability to think and communicate. In the support of one of our customers recently, we designed a simple to understand and use framework encompassing comprehensive detail as to their own specific values and behaviours, what they meant and the types and sorts of questions they should ask to inform and reassure them that the candidates sitting in front of them were aligned to the organisation and their expectations.
Additionally, we provided additional example follow up questions with which to probe candidates further. It is much harder for candidates relying on AI to therefore fake their responses.
There are some easier and quick wins that can be achieved in combatting the use of AI. These include cross referencing; validate their work history and/or claimed skillsets and experience by simply looking them up on LinkedIn, reviewing verified references and validations, etc. Look for mismatches within their writing style across different elements of the recruitment process. And if you really want to deter some candidates from using AI, clearly state that your organisation uses AI-detection tools and that you monitor misuse.
It should be said that there are many, candidates included, who do use AI ethically for things such as grammar checks or draft ideas. The goal needs to be to detect over-reliance or misrepresentation, not punish accessibility.
In conclusion, there is so much more for us to consider now with the rising use of AI and specifically with regards to recruitment. But we are working with it. Are you?
Kate Wilson is development manager at Osborne Thomas