Fed up of lazy AI use in recruitment

[DerelictWreck]

AIBU? Are others finding this?

We use an online recruitment system at work which is designed to reduce bias - we can only see how candidates have answered work based questions which are designed to assess their ability to do the job. We don't see any personal information or CVs.

As part of the process the candidates are told that use of AI will not help them and when they submit, they have to sign a declaration that it is their own work, not generative AI. Any AI work found will mean they are automatically dismissed from the process.

But over the last year this has so rapidly scaled that I'm now looking at over 50% of applications copying and pasting from AI chatbots. It's blatantly obvious, gives poor answers, massively slows down my recruitment processes, and is a complete waste of the candidates time as they get auto-rejected! I'm at the point where I'm also going to start blacklisting their names from future recruitment, and sharing said list with our sister companies (part of a large network).

I don't understand why they do it - using it to get ideas and editing it I get. But copying and pasting - how stupid do they think we are?!

They are very time consuming and so I’d have to be serious about the job to bother.

And that's exactly the point.

Exactly.

Surely that makes your job easier, 50% of candidates are telling you they are either lazy, not that interested, cant read instructions. Great in the bin, narrow down the pool, next!

Daft question, but if you have no data on them (so as to reduce bias), how can they subsequently be blacklisted? If all you get is the answers to their questions... what you going to do with that?

How are applicants supposed to know that you are an ’interesting, rewarding, values driven organisation that proactively invests in its people’? Presumably at application stage, all they’ve seen is your website and JD? And I’m sure it’s not news to you that… every organisation claims to be the above?

you’re assuming that the person who wants the job the most is the best person for the it. That’s not always the case.

The system has their details, I just don't have access to it when sifting.

So it’s ok for you to use AI but not the applicants?

We're pretty well known in our sector, and we seem to have a reputation for having a very positive working culture. A lot of applicants applicants tell us this. Many are also well known to us because they have worked with our partners or suppliers etc, so they have a sense of how we work and what we do.

And actually, wanting the job counts for a lot in our field, so while the person who wants the job the most won't always be the best fit, the person who doesn't really care about the job at all definitely won't be the right fit. We can teach a lot of the technical knowledge and skills that people might need, but we can't teach people how to share our values or have intrinsic motivation for the kind of work that we do, so we have to ensure that we are assessing these qualities effectively at the selection stage.

As I said above, if we get to the point where we can't recruit good people, we'll re-evaluate our processes to attract more applicants. Until such a time, we'll carry on doing what works for us, and anyone who doesn't like is is welcome to vote with their feet and apply elsewhere - we wouldn't be the right fit for them any more than they would be for us.

Comprehension problem?

The OP's post that you've quoted makes it quite clear that they are not using AI to screen candidates. They only use the AI tools to identify those who have submitted AI generated responses, ie to see if they match.

So they do use AI

To detect AI, yes. Don't see why that's unreasonable?

This is the fast track to losing good candidates. I don’t fill out applications like this, it’s hugely unnecessary and I found that big names don’t do this. I’m always more than happy to do skills based assessments at interview when usually they are better thought out and I know it’s not going be a waste of my time.

Great comment 👍🏻👍🏻👍🏻

Also, because of the way you described it, I am envisioning written answers as opposed to some sort of technical test. Apologies if I’ve got this wrong, but I don’t see how the former would help reduce bias.

Because how can we discriminate against any protected characteristic if we can't see them? Or have bias towards people like us, people with similar experience, people with privileged education etc if all we are seeing is how they would approach tasks that will be part of the role.

Op unfortunately everything is going to become ai soon, blandest load of crap.
School reports, children's actual work, teachers commenting, applications, lessons is all going to be ai.

The current generation of AI chatbots use a large language model.

Their responses are simply the most statistically likely combination of words for a given prompt.

Which would also explain you receiving lots of similar submissions.

So maybe the problem is that your questions are so clichéd that all your candidates are giving the most statistically probable response because you haven't given much scope for novelty.

It might be unnecessary for some roles but it is very useful for others. As I said above, the quality of our hires has improved since we switched to this system - we are getting better candidates, not worse ones, so the ones that are choosing not to bother aren't really a great loss.

How does that stop you being biased towards people who approach tasks the same way as you?

I was questioning the questions not blind applications.

A written answer will usually be quite telling even when explicit questions are asked. Particularly for more junior applications. The way language is used for example, the way challenges are approached, someone who is MC and has been coached how to sell themselves in written form since before their teens knows how to write a high quality answer. This is why there has been so much backlash against long form written applications for university and the like. If it’s not a particularly technical question, this in and of itself is barely more fair than a cover letter.

How does that stop you being biased towards people who approach tasks the same way as you?

Because applications aren't sifted/assessed just by me but by a panel, all independently. And then aggregate scores determine who comes to interview.

Can anyone who has taken umbrage with this approach tell me a fairer/better system which:

• is fair to candidates regardless of characteristics
• attracts good quality applicants
• isn't overly time consuming for either side?

I have no power to implement it but I'm curious what people thing the answer is, if it isn't cvs and it isn't this.

because to some people it depends on their version of what is considered cheating ,
as example : asking a friend for assistance or using ai to help give them an added advantage
plus when you have websites/books that basically do the same in suggesting how to improve x information then ai is the modern version of the old book version

for a start a company that is straight forward in what the candidate is being judged against as seen in early posts on this thread some people can match it exactly what is being asked by x and then still nothing from the company etc

I agree that recruiters should be transparent about how they will assess and shortlist applications.

However, in my experience, no matter how transparent we are about this, there will always be some candidates who believe that they have met the criteria whereas we don't agree with their assessment. This is much less of an issue for us now that we no longer use CVs because the questions are tailored more to what we actually want to know, rather than people's individual interpretations of how they meet a person specification.