AI is the promise for recruiters. It can take huge mountains of administrative work off their hands, improve the candidate experience, address the target group with the right tone-of-voice, select who are the top candidates for a vacancy and so on. Yet many recruitment professionals struggle with structurally deploying AI in their processes. What are the pitfalls to keep in mind? And more importantly, how do you make sure you don't tumble into it?
Quick summary
AI can bring huge benefits to recruiters, but without clear choices and agreements, pitfalls lurk. Many teams want to use AI, but find it difficult to build it structurally into their work process. Others get stuck because AI innovations are not linked to organisational goals or there are no guidelines within the team. It is also important to pour your own sauce over texts and be transparent about the use of AI during content creation. Ethical issues such as stereotyping, copyright and privacy also play a noteworthy role. By starting small, linking deployment to objectives, setting clear ground rules and always maintaining a personal touch and transparency, you will reap the benefits from AI without falling into its pitfalls.
Pitfall 1: wanting it, but not doing it
Recruiters are under high pressure: the labour market has been tighter than tight for years, the workload is high and managers are eager for candidates for their vacancies. Not surprisingly, the rise of AI tools and applications has been greeted with great enthusiasm. Surprisingly, the structural use of AI is still quite low, only 9%.
Our brain likes to stick to routines and familiar processes - even when there is a more efficient option. Initially, it takes energy to really build AI into your daily operations. Because where can you best deploy AI and which tool can help you do this? Finding the answer to this question takes time and that is exactly why it often lags behind. Before AI makes your working day more productive, you need to invest time in finding out and experimenting with various solutions.
How not to fall for it
- Start small and positive. Choose a task that was on your list anyway, but that you can do faster with AI. For example, summarising a digital job interview or optimising a job posting.
- Focus on one application and expand gradually. There are so many ways to use AI as a recruiter, that in your enthusiasm you might start working with 6 different tools. The risk in doing so is that none of them will really become integrated into your daily work and you will therefore get little benefit from them. Bring focus by starting with one application.
- Put the deployment of AI on the agenda of team meetings. Make sure it is a recurring topic where you share experiences and ideas and agree on concrete actions or pilots. That way, it will not remain just dreams, but a more efficient working day will become a reality!
Pitfall 2: AI as an end in itself
Hardly a day goes by that you don't hear or read something about AI. The hype is so great that the feeling arises: "We have to do something with this!" And applying AI in recruitment can indeed bring a lot, but without clear objectives, the impact is often limited. Guard against "using AI" becoming an end in itself, but a way to realise ambitions.
How not to fall for it
Whether you want to reduce the cost-per-hire or make the selection procedure more efficient, make sure you link the use of AI to recruitment or organisational objectives. Establish a baseline measurement and define KPIs before using the AI tool. Then schedule periodic evaluation moments and check whether the impact of the AI application is really as great as you had thought and hoped. Yes? Keep going! Falls short? Look at other solutions.
Pitfall 3: little trust in AI-generated texts
ChatGPT is the perfect digital colleague for all things text-related. Suddenly, anyone can write a flawless job ad or draft beautiful social media post. But... Your personal sauce is missing from these AI-generated texts - and people notice.
As a result, your texts will not work as well, because trust in an AI text is lower than in a text written by human hands. Trust in AI is particularly low in the Netherlands, according to research by KPMG. Only 33% of Dutch people say they trust AI. Texts that breathe the ChatGPT style are thus seen as less credible.
This is not just down to the risk of inaccuracies or incompleteness in AI texts. It is also in the effort put into a text. We know this phenomenon as The Labour Illusion. The more (visible) effort is put into something, the more people trust, appreciate and believe it. Even if that effort is not actually necessary. During an experiment with poems, you could clearly see this effect: people rated AI poems higher than regular poems. Until they knew that the AI poems were not written by humans. With that, the rating for the AI-generated poems plummeted. So how credible we find a text depends not only on its quality, but also on the effort put into it.
How not to fall for it
This does not mean that you should stop using ChatGPT for your texts. On the contrary, it is and will remain a great support in the writing process. But deploy it that way: as a help. When coming up with ideas, checking spelling, phrasing more rhythmically, but don't completely outsource the writing process. Pour your own sauce over it. As Désirée Battjes says in a Frankwatching article, "A bit of yourself, a bit of AI."
Another important tip: become a practiced prompt writer. The better your input to ChatGPT - for example, with articles with the desired tone-of-voice and context such as target group information - the better the output fits who you are as an employer.
Pitfall 4: no AI guidelines within the team
Deploying AI without agreements in your team is like driving without traffic rules: it goes well for a while, but sooner or later it goes wrong. Yet this is the practice: more than half of Dutch workers use AI in their work, but 69% say their employer has no policy for deploying AI. And this is not without consequences. According to the KPMG survey, more than half of AI users have made mistakes on occasion - due to the deployment of AI.
How not to fall for it
Have a conversation with each other about your AI use. What are you using it for? What are you running into? What do we think is possible and where do we draw the line? Based on those conclusions, draw up ground rules together. Think of guidelines like:
- New AI software is always first checked by colleague X for privacy and security.
- All AI texts are checked by a human. You remain responsible for checking the information and correcting errors.
- We never upload customer data or other personal information into AI tools.
- AI-generated images are OK, but we always mention this with the image.
A few simple agreements can avoid a lot of hassle and noise!
Pitfall 5: not thinking about ethical issues
AI brings with it numerous opportunities, which is great, but there is also a downside. There are ethical aspects that are important to keep in mind, such as:
- Stereotyping in your content: AI tools are trained by entering huge amounts of data. If stereotyping or bias is (unconsciously) present in the data the tool is trained with, you will see this reflected in your outcomes. WIRED magazine examined the popular video tool Sora and saw that women were shown mostly in support roles, such as flight attendant and receptionist, and rarely in managerial positions.
- Copyright: Recently, several issues have come to light where AI tools used copyrighted material for their training. For example, Meta was in the news this summer for allegedly using an illegal database for their AI chatbot that included texts by Dutch authors.
- Privacy: To support you in the best possible way, AI tools want to know everything about you. Some can watch or listen in via the camera and microphone in your laptop or phone, for example to see what you are looking at and translate this for you. Slowly we are moving towards a world of 'life logging' - everything you do is recorded - and that is not always pleasant. The episode 'The Entire History of You' from the series Black Mirror and the film (R)Evolution hold up a confrontational mirror when it comes to AI and privacy.
"If you ask ChatGPT to type up a job ad, texts with a lot of bias come out of it, because the system has been trained with existing texts on the internet that - subconsciously or not - contain biases." - Bas van de Haterd, AI and HR expert, in an article by HRMorgen
How not to fall for it
Being aware of these risks is already a big step. You are then more alert to them and may make different choices. For example, disabling advanced features for privacy reasons or not using a tool because it uses copyrighted data or the outcome could be stereotyping.
Pitfall 6: Loss of authenticity
One of the most important factors for a powerful employer brand is authenticity. At the same time, AI allows you to create all kinds of content these days - complete employer branding videos even - without using real people or images. Especially if you are not transparent about this, this can have major consequences for your employer image and the bond of trust with your target audience.
"Authenticity is so incredibly important in your employer branding and an AI-generated image is simply not real enough right now. People see and feel that. The same goes for texts. Of course, AI can contribute during content creation, for example for a quick post or in case of a limited budget, but try to focus on human, authentic content - that's how you will achieve the most." - Maarten Reeders, employer branding strategist
How not to fall for it
Maarten already mentions it: try to use real, own people and images in your content as much as possible. No AI office environment, but real images of the workplace where your future colleague will soon be walking around. Not AI colleagues, but the people you will soon meet at the coffee machine.
Your texts are also best equipped with your unique voice to avoid a '13 in a dozen' feeling and make a real connection with your target audience.
Finally, if you do choose to use AI-generated imagery, be honest about it. A small sidenote like 'video made with AI' is enough to inform your target audience and make sure they don't lose confidence in your organisation.
AI is the future, but in the right way
In short, AI brings a lot to recruitment. AI can boost your efficiency and effectiveness, as long as you avoid the pitfalls. Start small, link the use of AI to team goals, set AI guidelines, pour your own sauce over AI texts, take ethical issues into account and be transparent about the use of AI-generated content. This is how you get the most out of AI, without falling into any pitfalls!
