ChatGPT3.0, OpenAI's latest generative AI model trained to interact with users conversationally, is THE topic of conversation - you know it's universal when it's a dinner conversation topic! ChatGPT’s AI recent update does indeed take AI to a new level and convincingly writes code, answers questions, admits mistakes, challenges incorrect premises and rejects inappropriate requests. There are of course areas for improvement but even at a basic level, its usefulness is enormous and it will cause major disruption in multiple areas. Not least education, where a recent Independent article suggested that students are already using ChatGPT to create essays and answers (cheat?). The implications for recruiters will be no less significant.
On the positive side for recruiters, repetitive tasks like writing job descriptions, job adverts or even employer brand statements will be transformed into creative masterpieces. Interview questions can be constructed on the fly, all with model answers - you can even ask ChatGPT to come up with a rating scale. What about AI bias I hear you say. The step change in ChatGPT is that you can ask it to write in a neutral voice or if you want something with a bit more personality you can still get it to review its copy for bias. I wouldn’t be surprised if, by the end of the year, we are all reading ChatGPT prompt guides and getting certified.
There is also a negative side and not in the area, many assume. If ChatGPT can write the questions it can also write the answers. Any half-decent user will now use it to write the perfect cover letter and CV. There are already lots of articles and Reddit discussions on this.
The big question then is: How will recruiters differentiate between cover letters and CVs if they are all so good? LinkedIn remains filled with advice on how to write the perfect CV and cover letter and ChatGPT will do just that in under 60 seconds. Some firms are already seeing this - I heard recently from a Partner at a Magic Circle law firm, that the cover letters they get for early career applications are all of such a high standard there is no way to differentiate them. This has been true for many prestigious roles in places like Singapore, thanks to high levels of educational support and will now be the same for any knowledge-based role in any country.
If a computer can process information better than a human, we don’t need a human to do that processing. All the intelligence or aptitude tests in use today are based on processing information and deducing an answer, so whether you use verbal or numerical reasoning the format of the test is the same, as is what they measure, and this hasn’t changed for more than 30 years. One of the most well-known and still widely used in the legal sector is the Watson-Glaser test. Why would we measure a candidate’s capability to do something we know a widely available computer application can do better? Even worse, ChatGPT can figure out the answers and automatically enable a candidate to get over 70% on a mock version of the Watson Glaser test.
In a future of work enabled by ChatGPT and other computer-aided processing tools, the role of the worker will be creative, empathetic, ethical, and prioritised. These are human skills (or soft skills) that AI (in its current iteration) can’t replace. Job descriptions and therefore how we assess who is most suitable for them will need to value human skills higher than the traditional and coveted measures of intelligence. How you solve problems is just as important, if not more important, than simply identifying the answer.
As recruiters despair about deciding on how to short-list for an interview from the hundreds of brilliantly written application forms and CVs, the focus will shift rapidly to how technology can help. CV parsing applications will be meaningless because AI will be measuring AI. In fact, any text-based application (even SJTs) will be open to manipulation from ChatGPT.
There are two options, both of which will be needed. The first is a video assessment (which is really a video recording). This is clearly a human-centred approach. The problem is that attempts to automate the review of video interviews have been fraught with issues as one of the largest vendors in the market, HireVue, has discovered. If you can’t automate a 2-3 minute video assessment, then the advent of ChatGPT is going to force you into hours of watching video interviews instead of all the great things you were hoping to do with the free time it opened up from the other boring tasks.
The second option is to use psychometric assessments that are not text-based and use tasks and behavioural responses to determine human skills and the potential to be successful in a role. These types of assessments can’t be manipulated by AI and at the same time don’t require you to spend hours screening through the results. The better ones will enable you to self-configure what good looks like and match against that.
So before you get all excited about what ChatGPT will do to make your life easier, start planning how to future-proof your recruitment process to determine the difference between a real and an artificially enhanced application. Otherwise, you could be cancelling your Netflix subscription in favour of a different video application!
Check out our other article on ChatGPT further discussing the implications for recruitment: ChatGPT: a recruiter's friend or foe?
If you’re interested in discovering more about how hiring for potential can help you please download: The Playbook for CV-less hiring!