See the image below, found by someone 'somewhere' on the internet, and placed on LinkedIn stating that AI would simplify the whole process from searching for a job to recruiting the right candidate.
But is that the case?
LinkedIn is the 'Facebook Snapchat WhatsApp' equivalent for companies and industry these days. Originally it was a relation tool, but is has almost become a kind of Twitter / X variant where a business opinion can be unleashed upon an unexpecting world. For example: "The graph below illustrates that..."
Really? What does it actually show? The poster suggested that the whole process of searching for a job and selecting the right candidates could be so much easier using AI. But - again - is that the case?
Finding 'something somewhere' on the Internet is in itself not exactly an endorsement... That doesn't mean the research is faulty, but the data and the conclusion should always be questioned. Plucking a graph from the internet and declare the role of a recruiter dead might be a little premature. Models may and will help, but we should be very careful how we interpret the results.
The problem that I see is not the model, the problem is the data the model is fed.
AI versus Recruiter
1. Let's assume the model has originally been build by monitoring human recruiters, so it recognizes the candidates with a better chance on landing the job at the moment the model is build.
2. After the initial success the human recruiter will be taken out of the loop, and the model will be updated with its own success and failures. You run the risk of the model developing a bias by learning from its own output. If only people with blue sweaters are placed, then at a certain point it will only offer people with blue sweaters, because it thinks it has a better chance with those candidates. As only blue sweater wearing candidates are offered then the only successful candidates will be wearing blue sweaters (they all wear blue sweaters, after all).
3. Also, does the model take into account how long those candidates will stay with that company? How much they will contribute - or damage! - the projects they work on? How well will they cooperate with their (future) colleagues? How much knowledge, experience, attitude, or just plain enthusiasm will they bring with them? Can you even model that?
4. And finally: who decides if the model is successful or not? From a recruiter’s perspective the maximal number of hirings might be the quantifier, whilst to an employer it should be the max contributing value to the company as a whole.
Searchers
AI might help applicants LOOKING for a job. But where's the money in that?
Perhaps someone could build an AI search engine that applicants can use to find the best paid job with a simple press of the button. Who's going to develop that, how do you prove that it works, and who's going to pay for it? I see changes, but be aware of the limitations of the model. It won't be perfect, unless you include all data...
So... the AI should be integrated across the entire hiring journey and employee lifecycle, getting real-time data from all possible angles....
So, effectively, you want companies to disclose how long those employees work for them, you want them to quantify and qualify their value for those companies, and disclose that as well (as such a model would only work taking into account all employees and employers across all companies).
Good luck with that!
Still, there is a business case in there somewhere...
No comments:
Post a Comment