AI in Recruitment: How to Spot Cheating Candidates—and Hire with Confidence | Ep 247

Dr Steve Day: This may seem hard to believe. But one of my clients the other day shared that she was in a live interview interviewing a new remote worker virtual assistant. And in the live interview that person was sat typing away the questions that she was being asked. Then putting those into AI. And she was obviously reading off the screen what the AI was telling her to respond to the questions. This is the state of recruitment at the moment.

And today I want to share some of the other fantastic experiences that I\'ve had and that my clients have had. And also explain how we\'re combating this and actually finding new ways to use AI. To make our recruitment even more efficient and also avoid this nightmare situation. Where we\'re basically getting AIs applying for jobs through the medium of virtual assistants and remote workers. And this is valid for any type of recruitment. We have to be savvy about recruitment now.

We can\'t think that the answers we are hearing are actually coming from the person who is making the application. Today I\'m going to talk about how we\'re combating this. The advantages of adapting AI or adopting AI into your recruitment processes. So as I mentioned in the introduction, this is one of those things that when I heard it, I did not believe it. That somebody in a live interview was actually touch-typing the questions as they were being spoken to them. And then reading the response that was given by the AI back to the interviewer.

This to me was just like next-level cheating. But it just made the person look stupid and obviously they did not get the job. I thought this maybe is, you know, a one-off thing. But actually think about it. If I\'m hearing about it, it must be happening pretty frequently. Because, you know, I\'m just one little guy with a few clients. Whereas actually this is happening and becoming prolific. In fact, I believe in the industry. And the reason I know this is because we\'ve recently started using a piece of software. That allows us to track the mouse movements, the use of second screens, the use of AI tools and also the use of copying, pasting, etc.

And also it reads any responses in a text or verbal format. And actually assesses them for the likelihood that they were written by an AI. Really cool stuff. But what it has also brought to my attention is how many people are now cheating as they\'re going through application processes. Because let\'s put it frankly, what it is, it is cheating. They\'re cheating by using AI to answer questions for them. So we\'ve become really strict with the way we ask our questions. We explicitly are telling people not to use AI. And if we catch them doing it, we don\'t use the word catch.

We say, you know, if it is noted that you are using an AI tool, you will be rejected from the whole application process. And so we put it out there front and foremost. This is not acceptable behaviour. We even have a fair-usage honesty policy that comes up, which they have to sit and sign. If you like, confirm digitally sign before they start sitting the interview. To say they are not going to use AI at any point during the interview. But still people do it.

And so if you haven\'t got tools in place, it can be a real challenge. To actually know are these people genuinely who are applying for your job. So you\'re putting a lot of effort and time and money in behind. Actually going through and assessing for the potential fit for a role. Are they actually just using AI to cheat the system? And they arrive and have no ability to be able to do the job without the support of AI? And if you\'re hiring for a certain level of role, that\'s not acceptable.

You want people to come and bring expertise, not just their ability to use tools. And yes, they might be able to wing it. And you might say, well actually, if they can get the job done using AI, you know, isn\'t that what you need anyway? You just want them to be able to do the job well. Yay and nay. It depends. It depends if you\'re looking for basically someone who is just using a tool they don\'t really understand. To do a job they don\'t really understand.

The problem with that, I believe, is you\'re going to very quickly hit a ceiling. And you\'re going to get problems because they won\'t know how to undo the problems they cause. Or how to fix the issues that arrive or have to deal with exceptions. Because they don\'t have the experience to back up this stuff that they\'re actually doing. And it could actually cause huge problems if they don\'t understand what they\'re doing.

And they keep on going further and further down a rabbit hole. AI drags them down that rabbit hole. Thinking it\'s going the right way and you end up in a complete mess. And again, they have no idea. And I bet you those are the kind of people that are just going to disappear off the face of the earth when the problems arrive like that. When problems like that arrive, when you actually confront them about it and say, like what? How did this happen? And you actually ask them to sort of walk through their process and their thinking.

Those are the kind of people that unfortunately will probably just disappear. You never hear from them again. So how do we combat this? How do we actually create really cool recruitment processes? The genuinely brilliant people to shine and actually push away, reject or just identify those people who are trying to cheat the system. And we\'ve been using some cool tools recently.

There\'s an app called Testlify which I use. Which is quite expensive. But we actually, luckily, got it on a lifetime deal a while ago when it was on AppSumo. And so I get to play around with it for free now, which is really cool. But it\'s an excellent app. And if you\'ve got the incentive to actually go out and hire good people. You can get like a monthly fee for like, at the time recording, it\'s about $99 a month. And you can pay it for a month. And I believe that at the end of that month you can just unstop your subscription. But it will retain all of the setup that you\'ve got, the assessments and things in the back end. So next time you want to do your recruitment, you can continue where you left off. So it\'s really, really cool.

And if that\'s right, you check that out before you actually go up and sign up. And sort of spend hours setting everything up and then actually losing it all or anything. But I\'m pretty sure it can. I did ask their AI bot earlier just to make sure, before I recorded this episode. And it actually confirmed that you could do this. So therefore you can sort of use it for a month. Do your applications, do your recruitment process, and then switch off again. And use it next time you need it. So therefore, it becomes a relatively affordable app.

And when you realize what it can do and what it can save you in terms of, you know, poor applicants getting through. People just cheating the system, or actually the amount of time it saves you is just insane. So if you don\'t know, I\'ve been recruiting people for about nine years now. I\'ve recruited probably a couple of hundred virtual assistants, remote workers, admin assistants, systems administrators, marketing, social-media-type roles. And I\'ve been doing that for myself and also for our clients as well.

I\'ve also helped other business owners hire hundreds and hundreds more brilliant remote workers and virtual assistants for every role you can imagine in the company. And people who have gone on to become, you know, certified process consultants through our training have gone freelancing on their own. And through people who are now the office managers or even the operation managers within those businesses that we\'ve helped them hire for. So I\'ve got quite a lot of experience in helping not just myself, but other small business owners hire brilliant people.

However, with this advent of AI, we are seeing some real humdingers coming through. Like that example I gave in the beginning of this interview with someone who\'s so blatantly cheating. And I thought, as I said, it might be a one off. But recently I was just doing a recruitment drive for another one of my clients. And we started actually using some AI tools to help us to speed up our recruitment.

Part of it is about using things like ChatGPT and other tools to do big elimination rounds. So I\'ve created some really cool prompts for how to write using our Recruit Right process. I\'ve talked about Recruit Right in previous episodes. Recruit Right is our hiring system that we have through our platform, through our community, and we share it with our clients.

What that allows people to do is have a really structured four stage process where we\'re attracting, we\'re eliminating, we\'re testing and then we\'re selecting. And it basically just takes all of the thought and all the subjectivity out of a recruitment process. It allows you to give really objective assessments and be able to eliminate any number of applicants. And get down to that absolute short list that totally meets your company\'s values and your culture. And your assessment in terms of skills and competencies and capabilities and actually get them down.

So when you do the final interview, you\'re really interviewing the absolute cream of the cream, the cream of the crop. Sorry. So that\'s how we\'ve been doing recruiting for a long time. However, with this advent of AI, we\'ve noticed more and more people scoring really well at certain stages. But then just falling flat as soon as you get them to the late interview stage.

But then with people starting to actually cheat the system even with recorded interviews. So one of the key things and differences that we do compared to a lot of people I\'ve known who\'ve done recruitment in the past is that we don\'t interview anyone before we\'ve done a video interview with them. We basically ask them some questions and say, can you record a video and send that to us?

We maximize that video at five minutes and then we can watch that video at double speed. Two and a half minutes. We basically assess their English language, their ability to communicate effectively, their body language, their facial expressions, the ways they present themselves, the office environment, the background noise, and distractions. So many things we can assess from that little video interview.

And then we\'ve now been creating prompts to take that transcript and automatically do the assessment for us. Which has been a massive help with the speed of our recruitment, both for ourselves and our clients. We were amazed by how fast you can actually go through lots and lots of these video interviews. So we use software like ScreenPal, for example, which I\'ve talked again on this podcast about — screen recording software like ScreenPal.

We use that to record the interviews and then we can basically transcribe them automatically. Then we can get those transcripts, put them in ChatGPT using some clever prompts that we created. That then assesses the interviews based on all those things that I just discussed. That\'s all good, but the problem is that people are now getting AI to write their scripts for them and reading those scripts out.

They can have teleprompters and stuff, so it looks really natural when they\'re saying it. And actually with a bit of practice, you can get pretty good at reading off a script or even just memorizing the key points. You sound really good and you answer all the key points of the questions you\'ve been asked in a professional way. But it isn\'t actually your thoughts. It\'s not your experience coming through. It\'s just the AI that has told you what to say.

And so what we\'ve seen is that people are doing this and absolutely shining at the interviews or the video interview stage. But then we get them to a live interview and they just crumble because they didn\'t have the actual experience which they\'d made out that they had. So when you ask them to dig deeper into the things they talked about throughout the application process, they couldn\'t back it up.

And so it\'s just a waste of time. Waste their time, but massively waste our time, as well as recruiters and as business owners like myself. What I\'m trying to do now is to avoid all that. It\'s like, how can I be really sure that when somebody comes through to the point of us investing our time to set, you know, half an hour aside or whatever to do a live interview — the final shortlist — how can I be sure that the person that I\'m going to speak to and see on camera in that live interview scenario, for live that means for me a Zoom call.

But for many of our clients who have bricks and mortar type businesses like in the UK or on the ground, that could be a live interview as an in-person one. Whichever that is, it\'s a huge investment of time. So I want to be sure that when that person arrives, they are going to be the person I think they\'re going to be because of what they\'ve said on their applications.

And so now we\'re using AI to do a few things. The tool I talked about, Testlify, allows you to track things like mouse movements. If they use double screens, if they have AI tools switched on within their browser, it can detect if they are using copy and paste. It can detect the language that they use and whether it sounds like it’s come from an AI.

So that covers all the basic stuff about seeing if they are sort of cheating the system in plain obvious ways. But when it comes to the interviews, it can be a little bit more difficult because they\'re just recording a video. And then the AI software will then assess them in terms of their language and their communication and then grade their responses based on either the inbuilt AI-type criteria or you can give it your own scoring rubrics if you want.

And one of the cool things we\'ve done is to create ChatGPTs and then create structured interviews formatted in the correct way. So we can copy and paste them into this Testlify software. Then, literally, just post our job posting in. We\'ve set up the standardized way of the prompts to put out the form of the output. And we put in, for example, some really good recruitment training that I did. We put that as a source to then generate some cool questions. Then we just copy and paste that in. And then the AI within the recruitment software can use that for doing the interviews.

It basically talks them through like it\'s a live interview. That\'s awesome. So it basically says, gives the introduction, it\'ll say, hey, welcome to the interview, we\'re going to ask you some questions. If you\'ve got any problems, just ask whatever. Then it goes, right, first question, here you go. And then it pauses and allows the candidate to answer. The candidate can also say, hey, can you just repeat that? I didn\'t hear it.

And the AI will just talk to them like a natural person, like a human, and actually respond and answer any questions they\'ve got about the assessment as they\'re going through it. So it feels very much like a live interview. And what that allows you to do is to see how people are coping with a typical live interview scenario which is recorded in real time with a timer going down. It makes it much harder for that person to be utilizing AI because they can\'t move their mouse off the screen. They can\'t use two screens. You can lock it down really, really tight.

So it\'s very hard for them to be cheating in inverted commas. However, they could still have a second laptop doing stuff if they\'ve done this before and seen the system before. If it\'s the first time, it\'s very unlikely because they wouldn\'t know what\'s coming. The second time, okay, they could cheat the system a little bit, but even then you can watch what they\'re doing. It\'s recorded in real time. If their camera\'s off, just reject them. If the camera\'s on, it\'s hard for them to be looking elsewhere and doing whatever they\'re doing and still answer succinctly without looking and being obvious.

So if you do get people, or if we have got people that have done really, really well in a video interview when they weren\'t good at other bits, then we can actually dive in and have a human look at it and assess. Not sure about this, but it\'s this combination I found that\'s been really successful in using a cool app like Testlify to use those cool AI tracking tools but also using the AI software to do the interviewing as well. And under-gate AI software also grades the interviews as well.

That means we don\'t have to spend time looking through all the videos. It means we can actually “interview” more people and get a really good assessment of them based on some live interview-style questions which make people think on their feet rather than having time to type stuff out. You can time-limit all the questions, even multiple choice, long form, and short form questions. You can make it a much stricter, test-like environment which gives people far less opportunity to cheat.

With all the other bits I said before, and the other bells and whistles that sort of software comes with, you can actually eliminate the chance they can cheat at all really because they\'ve got to do it within the platform. And like I said, with this interview-style setup, it really makes it much harder, I believe, for them to get away with it. And I saw one the other day.

Their interview answers were really, really slick. But as she literally typed as the interview AI was speaking, she was typing away. The problem was every time she hit the keyboard, it interrupted the AI and you could see her getting really frustrated that she couldn\'t use the keyboard to get the answer she wanted. Then it just cut off. Funnily enough, she didn’t pass the interview in the end. She was doing really well until that point, then she just dropped off. I watched the video, and it was obvious why — because she was cheating the system, which is why she was so good.

Sadly, this is something I think we need to be aware of. I think that we need to be savvy when we are recruiting, more than ever. I think if you\'re using a recruitment agent, I would be asking them very specific questions about how they ensure the candidates that get through are not using AI. How much are they interviewing them one on one? What kind of assessments are they doing for that person outside of your typical questionnaire-type assessments?

If they\'re actually doing profiling assessments for personality types, for example, those are much harder to fake because personality, you could in effect if you had a job role and then you were given a personality test, you could probably get AI to help you answer it in the best way to fit the job role. But I would be asking, are they doing personality tests separate from the job role? Before they know what the role is, do they do the personality test? Because then there\'s nothing to cheat on.

You can\'t cheat a personality test. There\'s no right or wrong answer. You just are the way you are, and then you are a good fit for a role or not. So yeah, there are things like that you could ask. I\'d be really just thinking hard if you\'re going to use an agency. If you\'re not going to take on the recruitment yourself, make sure the agents you\'re using are aware of all this stuff.

If you are going to do the recruitment yourself, then use multiple touchpoints and tools that allow you to assess if that person used AI. Even a really simple thing that you can do is give them the opportunity to write long form answers. So give them an open question and say write a long form answer. You can really quickly create a GPT, a ChatGPT prompt that will then allow you to assess that response against a job posting.

So we\'ve done this many times. We\'ve got some really cool prompts doing this now. You get your job posting, you create a prompt that basically says, you\'re a recruitment specialist. You\'re going to assess a candidate\'s response based on a job posting. I\'m going to upload the job posting and the candidate response and I want you to give me a grade one out of ten to see how they fit that particular job.

If you\'ve got multiple candidates, do 10 at the same time. Say, look at all these 10 and grade them 1 to 10 and compare them against each other. Give me a summary at the end. And you can also, if you want to take this one step further, create a rubric scoring rubric. It can be used against any number of assessments so you get more consistency. You can get ChatGPT to help you create that. So the whole process would be: set up the prompt like I described before. Give it a role, the inputs, the rules it needs to follow, including the scoring rubric. Tell it how you want the output to appear. That\'s basically prompt 101.

Then ask ChatGPT to create a scoring rubric that you can use for any future job role going forward. It can then create a generic scoring rubric that you can reuse. You\'ve got a cool, reusable scoring system where you just upload the job posting and candidate responses. It uses the baseline rubric you\'ve already created. Now you\'ve got a way to assess your long-form and even interviews as well, because you can get the transcript from the interview, like I described earlier, put that in, and it can do the same thing. That\'s a really cool way of doing it if you\'ve not got fancy software.

The other thing to do is to ask the AI, was this written by an AI? Or you can use a tool like Grammarly Pro, for example, which is a brilliant AI-based software used to help you write grammar and check spelling. I\'ve used it for years being dyslexic. It\'s been an absolute lifesaver for me for many years. They now have a feature where you can paste text in and it tells you the percentage written by AI. You can cheat it though.

For example, any of my emails actually written by AI — although we’ve spent many, many hours training it to sound like my voice — now don’t get picked up as AI at all. But that’s difficult. You need to be savvy about it. And if somebody’s that good at AI, then I’d probably want to hire them anyway because they’d be a massive asset to the company.

Anyway, we digress. The point is that there are ways to assess if somebody is cheating with AI. If you’re not going to use fancy software like Testlify, then make sure you’re using the Recruit Right hiring process. Go back in my episodes and search for Recruit Right. I’ve got loads of episodes which talk through that entire process really succinctly. The one we’re looking at here is the long-form questions of the test stage. Ask them to write some long-form answers in response to questions around culture and experience.

Tell them to describe a time when they did XYZ and then put it through an AI testing tool like Grammarly and score it as well using AI. And now you’ve got a pretty slick system for free. Whereas if you’re using something like Testlify, it’ll cost about $100 a month. I think that’s it really for today. It was more just a “wow, I can’t believe this is happening.” I can’t believe people are sat in live interviews using AI to cheat their way through. That is just next-level. Wow.

Did not think, did not see that coming. But actually I should have done. It’s been an eye-opener to see how good the tools are now that help us combat people using amazing tools to cheat the system. It’s just an arms race, basically. I don’t know if you’ve read The Blind Watchmaker by Richard Dawkins. It’s all about the arms race between natural selection. He talks about the cheetah being just that little bit faster than the antelope. The only reason the antelope’s that fast is because the cheetah’s that fast.

The only reason the cheetah’s that fast is because the antelope is that fast. They basically naturally select themselves to be faster and faster. They’re the pinnacle of speed animals of the world. Same with this. AI isn’t just a thing. It’s what the thing can do for us or against us. As business owners, we’ve got to be savvy. People are going to use AI to benefit us and also to cheat us. We have to get that balance right.

The tools to beat the cheaters will progress just as fast — normally one step behind — the cheaters are probably one step ahead. Then we catch up and catch up, and hopefully keep a roughly level playing field as we go forward, as the world of AI explodes. So just be mindful that when you’re doing recruitment, don’t believe everything you hear. Don’t believe everything you read.

Make sure you are being savvy when you’re going through the selection process so you don’t waste time hiring people that are basically AI drones just repeating whatever AI is saying. Make sure you always, always do live interviews. And in those live interviews, ask the person not to touch their keyboard. Ask them to look at the camera throughout and not read anything off-screen.

Say at the beginning, if we think you are looking at a screen or typing on the keyboard during this interview, we’ll take that as cheating. Explain why so it doesn’t feel like a personal attack. Say, we’ve had lots of problems in the past with people cheating on interviews, and now we ask people to sit and not use anything on the computer while doing the interview.

Just so we can be confident that the experience, expertise, and competence you’re displaying is your own. Good people will thank you for it because it gives them an opportunity to shine and show how good they are. They’ll stand out against people trying to cheat the system. Don’t be afraid to talk about these things. I think if you are honest about them, and you explain the reason why, honest people will not have an issue with it.

The only people who have an issue are those actually trying to get one over on us or whoever they’re trying to cheat. So that’s it. I hope you have a fantastic day. If you found this episode helpful, please share it with your colleagues, your friends, your loved ones — anyone who might find it useful — and please do hit subscribe so you don’t miss out on future episodes. Thank you very much indeed. Until next time, see you again. Bye.

AI in Recruitment: How to Spot Cheating Candidates—and Hire with Confidence | Ep 247

The key is staying AI-savvy and one step ahead:

KEY TAKEAWAYS: How AI in Recruitment Is Changing the Game

BEST MOMENTS: The Real-World Impact of AI in Recruitment

TIMESTAMPED OVERVIEW

Episode Transcript

VALUABLE RESOURCES

LINKS TO CONNECT WITH THE HOST

ABOUT THE HOST

For more articles related to leveraging AI in recruitment and hiring virtual assistants, you may also like:

SYS 067: Automate Your Customer Onboarding – Interview with Joe McNee

SYS 199: Why We Use Fathom AI Notetaker: Pro Tips to Maximise Value and Minimise Cost

SYS 179: Simplifying App Development for Business Owners – Interview with Norman Ortiz, Founder of App Studio

Discover How Small Business Owners Leverage The SixR System To Achieve Time Freedom In Under 90 Days

How Small Business Owners
Leverage The SixR System
To Achieve Time Freedom
In Under 90 Days

Free On-Demand Masterclass