As an interviewer, I'm seeing a huge increase in proportion of candidates cheating surreptitiously during video interviews. And it's becoming difficult to suspect any wrong-doing unless you're very watchful by looking for academic responses to questions.
Why would anyone encourage building such a tool, I can't fathom.
It's pretty simple - people need to eat (and fulfill other basic needs, of course), to eat they need jobs, to get jobs they need to pass the interview. The hiring process in a lot of industries is heavily gamed at this point, to the point that not cheating is basically an automatic fail. So, if you want to eat, you cheat.
> The hiring process in a lot of industries is heavily gamed at this point, to the point that not cheating is basically an automatic fail.
This sound a bit of "thief thinks everyone steals". Interview preparation is normal and common but I don't think cheating is. May depend on the location of course.
I can totally understand thinking this way out of desperation, and being lulled into thinking it’s this simple, but it seems short sighted with hidden complexities. First of all, it’s risky. If you get caught, you don’t eat, and it could follow you and prevent you from even getting in the door elsewhere. Companies are always going to be watching for cheaters, they are always going to have more visibility than you into what interviewees are doing, and they are always going to have more resources. Even if you do cheat and get hired, it quickly becomes obvious that you’re unqualified and can’t do what you claimed, and even if you don’t lose your job, you’re less likely to get promoted. Being lazy and amoral about interviews seems like a trap people set for themselves.
The good news is that a lot of companies are starting to allow AI during the interviews, and suddenly it’s not cheating. But of course that means you need to be good at using AI and interviewing and programming, you won’t be able to cheat and rely on the AI to do your talking for you.
Doing whatever it takes to get the foot in the door may be encouraged, but only to a point and I think out and out cheating is probably crossing a line... As would murder, arson etc. etc.
If cheating means asking someone in the company you're interviewing for a peek at what will be asked then great. In my book that's using leverage.
Reviewing previously posted interview tests is probably recommended.
Hooking up a copilot to answer interview questions for you in real time is probably less so.
Probably you've been out of the getting hired game but I had a glimpse of it last year: absolutely terrible.
When I started you'd send a mail to the company directly about a position, you'd go to the office, have a short interview, meet the team and they'll let you know. That's it.
Now it's 2 rounds of HR bs, 3 layers of tech interviews, then meet the CEO/CTO/etc. And then references and then a final "chat". And you still can get ghosted at literally any step, even at the final cozy chat, just because of "vibes".
And throw in companies sending you leetcode even before talking to you and you can see why one would want to get through the bs.
I still stand about my favourite approach for tech jobs: intro and tech chat (1-2h) about your resume, what you'll be doing and anything you might have questions about (no challenges or stupid riddles). Then, if everything goes smoothly, you get a 2 weeks contract and you are in probation. If everything goes well, you get another contract for 3-6 months (up to you to accept or not) and then you get converted to permanent if everything went well for both parties.
I actually like your idea of a probationary hire, but you can see this is just an even longer extended interview, right? If companies were to adopt this model en masse, they would over-hire and then drop most people after the first 2 weeks, and you’d be out looking for another job, having wasted even more time than 5 rounds of interviews, and being unable to interview for multiple jobs at the same time.
Software interviews and hiring have definitely changed over time, and I know it’s harder right now, but I think we’re seeing the past with slightly rose-tinted glasses here. It was never only just one short interview, there were applications and emails and phone screens. In my career, I’ve always had multiple interviews and technical discussions during job applications, even back in the 90s. Getting hired, for me, has always taken several weeks end to end, if not longer.
There are a bunch of reasons interviews are getting harder, and people trying to game the system and trying to cheat are one of them, a big one. Think about it from the company’s perspective: what would you do if the volume of applications you got started far exceeding the number of positions available, and an increasing percentage of the applications you got were people unqualified for the positions but adept at pretending? More face time vetting before hiring seems like the only reasonable answer.
Other reasons why interviews are getting harder is that software jobs are more competitive now, and possibly relative pay has gone up. If interviewing was easier back in the day (and I agree that it was), it’s because there wasn’t as much competition.
I am old and thankfully out of the getting hired game. I was cleaning out some files (paper!) recently and ran across correspondence from old job searches. As you said, single visit and decision. I was also struck by the number of letters from companies thanking me for my resume and politely telling me they were passing but would keep me in mind for future openings. It was not uncommon to receive a letter directly from the hiring manager thanking me for coming to an interview.
A two week probation means that nearly all candidates will need to quit their current job to do the probation which seems unlikely to be popular with candidates
I won't use it, but I do see it as somewhat symmetric. If the interviewers are using AI or expecting you to use AI for these tasks once you're on the job, then it doesn't seem completely immoral.
Get ready to start having some fun in your interviews. Start including things like redirection of focus through general statements, unrelated (and false) trivia, and misleading suggestions in your interview questions. Most of the humans you'd like to hire will ignore those or ask you about them.
Many LLMs will be derailed into giving entertainingly wrong answers:
> unless you're very watchful by looking for academic responses to questions
I've noticed that a lot of the supposed hallmarks of "AI slop writing" (em-dashes, certain multisyllabic words, frequent use of metaphor) are also just hallmarks of professional and academic writing. (It's true that some of them are clichés, of course.)
It seems like most efforts to instruct people on how to "fight back against AI writing" effectively instruct them to punish highly literate people as well.
I think it's often still possible to tell human writing that uses some of the same tropes or vocabulary apart from AI writing, but it's very vibes-based. I've yet to see specific guidance or characterizations of AI writing that won't also flag journalists, academics, and many random geeks.
Honestly, why would you care? IF, and this is a big if, you are confident your interview process accurately assesses the abilities of candidates to carry out the role, then why would LLM assistance even matter? Are they not going to be allowed to use LLMs on the job?
This faux-outrage is just showing how broken the whole hiring process is in tech.
Stop giving people puzzles and just talk to them. If you're unable to evaluate if someone's a good fit for a role then you either need to learn more about effective interviewing, learn more about the role, or find someone else who is good at hiring/interviewing.
Indeed, I am sympathetic to the author in this situation because I think open-source is important, but I don't approve of this software and don't want to affiliate with it by even starring it on GitHub.
Not really sure what I can do for the author but say "that sucks, bro".
As an interviewer, I'm seeing a huge increase in proportion of candidates cheating surreptitiously during video interviews. And it's becoming difficult to suspect any wrong-doing unless you're very watchful by looking for academic responses to questions.
Why would anyone encourage building such a tool, I can't fathom.
When interviewers use LLM gargbage to filter out participants, expect candidates to do the same.
I don't use LLMs to filter out participants and I expect candidates not to use LLMs to cheat.
11 replies →
It's pretty simple - people need to eat (and fulfill other basic needs, of course), to eat they need jobs, to get jobs they need to pass the interview. The hiring process in a lot of industries is heavily gamed at this point, to the point that not cheating is basically an automatic fail. So, if you want to eat, you cheat.
> The hiring process in a lot of industries is heavily gamed at this point, to the point that not cheating is basically an automatic fail.
This sound a bit of "thief thinks everyone steals". Interview preparation is normal and common but I don't think cheating is. May depend on the location of course.
1 reply →
> if you want to eat, you cheat.
I can totally understand thinking this way out of desperation, and being lulled into thinking it’s this simple, but it seems short sighted with hidden complexities. First of all, it’s risky. If you get caught, you don’t eat, and it could follow you and prevent you from even getting in the door elsewhere. Companies are always going to be watching for cheaters, they are always going to have more visibility than you into what interviewees are doing, and they are always going to have more resources. Even if you do cheat and get hired, it quickly becomes obvious that you’re unqualified and can’t do what you claimed, and even if you don’t lose your job, you’re less likely to get promoted. Being lazy and amoral about interviews seems like a trap people set for themselves.
The good news is that a lot of companies are starting to allow AI during the interviews, and suddenly it’s not cheating. But of course that means you need to be good at using AI and interviewing and programming, you won’t be able to cheat and rely on the AI to do your talking for you.
Doing whatever it takes to get the foot in the door may be encouraged, but only to a point and I think out and out cheating is probably crossing a line... As would murder, arson etc. etc.
If cheating means asking someone in the company you're interviewing for a peek at what will be asked then great. In my book that's using leverage.
Reviewing previously posted interview tests is probably recommended.
Hooking up a copilot to answer interview questions for you in real time is probably less so.
5 replies →
Probably you've been out of the getting hired game but I had a glimpse of it last year: absolutely terrible.
When I started you'd send a mail to the company directly about a position, you'd go to the office, have a short interview, meet the team and they'll let you know. That's it.
Now it's 2 rounds of HR bs, 3 layers of tech interviews, then meet the CEO/CTO/etc. And then references and then a final "chat". And you still can get ghosted at literally any step, even at the final cozy chat, just because of "vibes".
And throw in companies sending you leetcode even before talking to you and you can see why one would want to get through the bs.
I still stand about my favourite approach for tech jobs: intro and tech chat (1-2h) about your resume, what you'll be doing and anything you might have questions about (no challenges or stupid riddles). Then, if everything goes smoothly, you get a 2 weeks contract and you are in probation. If everything goes well, you get another contract for 3-6 months (up to you to accept or not) and then you get converted to permanent if everything went well for both parties.
I actually like your idea of a probationary hire, but you can see this is just an even longer extended interview, right? If companies were to adopt this model en masse, they would over-hire and then drop most people after the first 2 weeks, and you’d be out looking for another job, having wasted even more time than 5 rounds of interviews, and being unable to interview for multiple jobs at the same time.
Software interviews and hiring have definitely changed over time, and I know it’s harder right now, but I think we’re seeing the past with slightly rose-tinted glasses here. It was never only just one short interview, there were applications and emails and phone screens. In my career, I’ve always had multiple interviews and technical discussions during job applications, even back in the 90s. Getting hired, for me, has always taken several weeks end to end, if not longer.
There are a bunch of reasons interviews are getting harder, and people trying to game the system and trying to cheat are one of them, a big one. Think about it from the company’s perspective: what would you do if the volume of applications you got started far exceeding the number of positions available, and an increasing percentage of the applications you got were people unqualified for the positions but adept at pretending? More face time vetting before hiring seems like the only reasonable answer.
Other reasons why interviews are getting harder is that software jobs are more competitive now, and possibly relative pay has gone up. If interviewing was easier back in the day (and I agree that it was), it’s because there wasn’t as much competition.
2 replies →
I am old and thankfully out of the getting hired game. I was cleaning out some files (paper!) recently and ran across correspondence from old job searches. As you said, single visit and decision. I was also struck by the number of letters from companies thanking me for my resume and politely telling me they were passing but would keep me in mind for future openings. It was not uncommon to receive a letter directly from the hiring manager thanking me for coming to an interview.
A two week probation means that nearly all candidates will need to quit their current job to do the probation which seems unlikely to be popular with candidates
2 replies →
I won't use it, but I do see it as somewhat symmetric. If the interviewers are using AI or expecting you to use AI for these tasks once you're on the job, then it doesn't seem completely immoral.
That's assuming all interviewers are using AI. And if it's not immoral, why do it surreptitiously?
1 reply →
there was already a paid and closed source application, i didnt create anything new
Get ready to start having some fun in your interviews. Start including things like redirection of focus through general statements, unrelated (and false) trivia, and misleading suggestions in your interview questions. Most of the humans you'd like to hire will ignore those or ask you about them.
Many LLMs will be derailed into giving entertainingly wrong answers:
https://arxiv.org/pdf/2503.01781
> unless you're very watchful by looking for academic responses to questions
I've noticed that a lot of the supposed hallmarks of "AI slop writing" (em-dashes, certain multisyllabic words, frequent use of metaphor) are also just hallmarks of professional and academic writing. (It's true that some of them are clichés, of course.)
It seems like most efforts to instruct people on how to "fight back against AI writing" effectively instruct them to punish highly literate people as well.
I think it's often still possible to tell human writing that uses some of the same tropes or vocabulary apart from AI writing, but it's very vibes-based. I've yet to see specific guidance or characterizations of AI writing that won't also flag journalists, academics, and many random geeks.
Honestly, why would you care? IF, and this is a big if, you are confident your interview process accurately assesses the abilities of candidates to carry out the role, then why would LLM assistance even matter? Are they not going to be allowed to use LLMs on the job?
This faux-outrage is just showing how broken the whole hiring process is in tech.
Stop giving people puzzles and just talk to them. If you're unable to evaluate if someone's a good fit for a role then you either need to learn more about effective interviewing, learn more about the role, or find someone else who is good at hiring/interviewing.
This has all been a long time coming.
Indeed, I am sympathetic to the author in this situation because I think open-source is important, but I don't approve of this software and don't want to affiliate with it by even starring it on GitHub.
Not really sure what I can do for the author but say "that sucks, bro".
[flagged]
If a question you are asking in an interview can be answered immediately by an AI, then why hire for that position in the first place?