It's a good point haha :-) Maybe caused by the fact I've done a fair bit of recruiting (although not at large companies) some years and sometimes drown a bit. I can relate somehow.
I'd bet my hat that most of the recruiters never even saw a single resume that came from the AI, and that the vast, vast majority were immediately filtered by the ATS.
Doesn't even have to be anything big, just that there aren't enough keywords, or that there were gaps between employment that were too large, etc.
The ATS would then rank the applications and score them, and HR just looks at the top 15 or something and picks the top 5, whittles them down to 2-3.
They're not spending 8 hours a day reading resumes.
That is survivorship bias though. Maybe you had like 5 opportunities and got three jobs. If they hadn't worked out, finding the 6th would be way harder.
For those pigeons where there are no holes, it instead seems impossible.
My success rate was absolutely terrible during the dotcom crash. After that I've increasingly relied on intermediary recruiters to find me work. They're clearly much better at that than I am. So these days I get asked and I reject them.
maybe you aren't stretching far enough. if you apply for jobs you are absolutely qualified for, obviously you'll get them. it's probably worth some effort to get a ~30% more difficult job with 30% more pay for most people.
I never had much problem with applications in the last, but in the last job search it was futile. I’m still just now getting rejections from jobs I applied to months ago which seem a perfect fit for my resume.
I even applied to a place I worked a ~5 years ago, which I got via cold application. Apparently I’m not good enough anymore because I couldn’t get past the filters.
I finally found a job through a referral. Oddly, it was working with tech I have basically no experience in and probably had the most awkward interview out of all the ones I ended up doing.
It’s a pretty grinding process. I’ve been fortunate in the past, staying with 2 companies over the span of 15+ years and I am now having to job hunt for the first time since 2005.
I found a local contracting opportunity that sounded really interesting and something I could probably be a good fit for. It was a small company, 4 people, so I actually emailed the president directly with a well thought out introduction and explaining why I thought I would be a good fit for the position. Radio silence…
So, even the old adage of ‘hit the bricks and knock on doors’ has gone the way of the dodo.
I was legitimately considering going to their office but it being a small company, in the medical field, and uncertainties about how individuals might react with COVID still present…I opted for a personal email to the owner of the company. Maybe I should have opted for the ‘cold call’ approach and just went to their office but I erred on the side of caution/ decorum.
This is the future. Every unemployed person will have an AI scanning all the job listings. They can send off 5k applications a day. Conversely, every open position will receive 100k applications, almost all of which will have fictional resumés and perfectly tailored cover letters. These will be fed into the company AI, which unavoidably has to make hiring decisions.
The actual hire will be someone the hiring manager knows from college.
But this use case shows that the rate of success is low when using ai. So if anything it would indicate that using ai for job applications is not a good idea. If i see ai spam in a job applicant’s email i am rejecting them instantly.
I'm not sure about that, it probably depends on the position. In my case, If I send out 10 resumes, I get ignored by 3 potential employers, 3 will politely write back that "after careful review...", but 4 will invite me for the interview and at least one make an offer (usually more than one). So at least in my case, using "AI" (or: automatically generated resumes) is actually more of a risk than advantage.
How many of those 20 interviews do you suppose were at great places where the person would enjoy working? They are evidently the places with the worst resume screening (or, alternatively, the most desperate hiring managers). Maybe there’s a reason they’ll interview anybody who tosses in a crappy AI-generated resume.
you're assuming that the Mk 1 Eyeball does a better job of sorting through candidates than "crappy AI" resume.
And history has clearly proved that's just not the case.
Most resumes follow a template and are scanned for keywords and action-phrases. ATS software isn't smarter than that, outside of maybe making some guesses about you based on time worked at each job or if you're lacking attention to detail by not ending each bulletpoint with a period.
Lucky you. I know someone that has been trying very hard to get employed again but is now for two years without a single positive response. Sure they're doing 'something wrong', no doubt or their resume might be off-putting but your experience isn't necessarily the experience that everybody else has.
anecdotes aren't data, and the market has shifted a lot since 2020.
remote is here to stay, and AI will be eating a lot of jobs.
call my bluff: throw out 3 resumes this week and see what sticks. How many come with the same level of salary -- it's dropping, aggressively, across the market -- or more?
I'm well aware, but the article we're responding to is an anecdote as well.
> remote is here to stay
thankfully, it's been great for me
> call my bluff: throw out 3 resumes this week and see what sticks. How many come with the same level of salary -- it's dropping, aggressively, across the market -- or more?
I don't see the need to do this, I'm happy at my job and don't even know who you are. You're probably right, why would I waste my time?
You don't see AI spam. All I have to add to the AI prompt is "use the style of Matt Levine", "patio11", "Hemmingway", "a 10 year old on crack" or "Donald Trump" and the letter looks nothing like what you think AI writing looks like.
It just came to my mind that, for all the talk about cryptocurrency being "funny money", this could actually be an interesting use case for utility tokens.
"Hey, do you want to apply to the position? Tell me how many Vitae (tm) tokens you want to pay for it."
It doesn't need to be worth a lot, it just needs to be something.
You make it inflationary. You reset the supply every year. You make it in a way that it is super easy to add companies to the list of approved minters, and let them mint whenever they want so that they can give to their prospects.
The token itself has no monetary value. I'm literally meaning it as a way to be an utility token, which is just another way to say "funny money".
The idea is just to have a way to avoid spam applicants. There will be no point in hoarding it, unless you are just trying to stop everyone else from ever submitting job applications.
> unless you are just trying to stop everyone else from ever submitting job applications
Sure. Less of them means more success for me. Or for my people, i.e. we here at Stanford hoard it in order to ensure the success of Stanford grads.
In principle I'm alright with the idea -- force people to limit themselves to 25 applications, stuff you really want, keep down the noise -- but what happens if people don't get a bite at 25? what are you forcing the hiring orgs to do?
No, less of them means that recruiters will not receive any applications in that channel. If recruiters stop caring about that channel, there is nothing stopping them from using or even creating a different one.
The micropayment problem is not new, and there's no indication that the Ethereum-based tokens have solved them, especially not with transaction fees that high. If you're talking about a non-decentralized solution, then as the sibling comment argued it boils down to an entry in a database and that's trivial anyway.
We are talking about an application with literal zero monetary value. No one will attack a test net to be able to mint themselves a token meant to be used solely for priority-rating in a process queue.
We can't have it both ways: either paying a token is a disincentive to spam job postings, which means the token actually has monetary value that's higher than zero; or it's not, then it's useless.
I mentioning transferring small dollar amounts just to show that it is not feasible to do it with regular money transfers.
> (either) the token actually has monetary value that's higher than zero; or it's not, then it's useless.
Are you familiar with Germany's Pfand system?
You go to a beer garden, get your pint which costs (say) 5€. The bartender asks you for 7€, meaning a 2€ "pfand". You get your beer and a physical, plastic token.
When you are done with your beer, you go back to the bar and if you return your glass with the token, you get your 2€ back. The key thing is that you need to bring both. Lost the token? No cash back for you, might as well take the glass home. Lost/broke the glass? The token is worthless.
The point here is that the token, by itself, has no value. It's sole purpose is to be a unit of accounting.
In your example, the 2€ deposit works as an incentive to bring back the glass. Without this incentive, the token on its own is useless. The token merely serves to prevent some side-effect of the deposit (e.g. people could try to return glasses from other patrons or other bars).
Similarly, having job tokens independent of any kind of deposit or incentive will not prevent anyone from spamming applications.
Prior to the 18th century reforms, British army commissions (officer jobs) were explicitly bought and sold.
> It doesn't need to be worth a lot, it just needs to be something.
Market pricing, though. I can see the rate for being considered for lucrative tech jobs climbing into tens of thousands of dollars (or one or two paycheques).
But we are not talking buying the job, we are talking about buying the spot in an application queue. The recruiter would still reject the applicant regardless of how much they paid.
And the AI will be sending rejection/acceptance emails to the candidate which in return will send an AI/generated response back, which will result in an AI giving an interview to someone who created an AI to do said interview..:
So job interviews will incorporate Voight-Kampff test-style questions?
"The tortoise lies there, its belly baking in the hot sun, beating its legs, trying to turn itself over, but it cannot do so without your help. You are not helping. Why?"
It doesn't really assist because it will just become an arms race between AI applications and AI filtering, which some companies have already started using.
Technology makes it easier to be more anonymous, superficial, and lazy on both sides of the equation. Each side will gain a small temporary advantage with every advance in AI and the end result will be that EVERYONE will be worse off than before.
That's because the job application process will be worse, people will feel more isolated, and you'll still need to spend just as much time as before if not more learning all these arcane tools.
Technology is truly becoming a disease. We've taken something good and used too much of it like sugar. Everything has a happy medium, and technology used with moderation could have been something good, but its use is being pushed to its logical conclusion through these prisoner's dilemma-type situations and is resulting in a degradation of society.
AI only highlights how ridiculous the hiring process is. A program that generates likely word sequences determines whether or not you get the job? What is even the point of that? The company looking for employees would be better off looking for people that are actually good at the job they are supposed to do there. (Unless that job is generating likely word sequences -- then they should buy AI instead of hiring people).
Without AI, such a process favors potential employees that are good at writing likely word sequences themselves, and not those that are good at actually dpoing the job.
You just pointed out how RIDICULOUS technology is in this aspect. The arms race started before AI -- if people had no computers to submit resumes, they might actually have to go in and TALK to someone...
"If people had no computers..." Most of the people here are tech workers our entire industry would not exist then. Would you have us all work in the typing pool copying out letters on typewriters to snail mail?
> The company looking for employees would be better off looking for people that are actually good at the job they are supposed to do there.
That's how most companies do it, even today. There is a special subset of companies that overautomate the hiring process, but it's not most of them (outside of SV-style companies, anyway).
It’s as if suddenly a significant number of ai people are hell bent on spam, ip theft, putting people out of jobs, spreading fear, manipulating and gaslighting folks, and so on. Almost as if after decades of irrelevance and frustration at memes about if then else “ai” attempts they seek revenge.
The tone was set by their thought leaders, in particular sam altman and the openai clique.
Taking a tech that was supposed to be a force for good and turning it into a toxic charade is quite the achievement. But gradually people are starting to push back.
True...in a world where resources are slim and the world is getting more crowded, it's hard to believe that the inventors of AI wouldn't have thought that their technology would be used for a whole truckload of nefarious purposes.
People like Sam Altman just go after short-term profit because they can. They are psychopathic.
and I don't care if large corps with automated ATS systems are mildly inconvenienced.
Statistically, most people weren't getting hired anyway with standard, non-AI ATS systems + the Mk1 Human Eyeball.
And of the hires, if you define successful hire as 2+ years in the role, most new hires fail at a rate between 50-80%[1]. Clearly not having AI is a failure. It might create tons of spam, but there is already a ton of spam, and it's miserable for everyone.
> I don't care if large corps with automated ATS systems are mildly inconvenienced.
Neither do I. However, this use of AI harms more than just them -- it harms all other job-seekers as well. It looks to me like doing this just accelerates the race to the bottom for all.
I dunno, if the success rate is only half a percent it's not selling it to me.
Mind you, I've never applied for a job myself, it's all gone through recruiters. This shows that that's going to continue, it's real people knowing real people, networks etc.
The price (starting at $99) is a huge turn-off. I don't have that much money to spend at all; I don't have a job, for crying out loud. I'd rather sign a contract to give away the first $250 or so of my income. Hell, even make it $500. Just don't make me pre-pay for the chance to win a job - I cannot afford to gamble on this.
I found myself in a similar position (unemployed) after I left my cozy (but soul-sucking) corporate job last year and [disclosure: ] I built a somewhat similar tool but I decided against automating the application process.
Instead of focusing on quantity - number of applications - I decided to focus on quality - better resumes that emphasise the candidates skills that align with the job ad.
And instead of applying to all job ads, my service analyses a candidates resume and finds positions that closely match the candidate experience.
I find interesting that he only got 20 interviews after the ai sent 5k applications and same 20 interviews after applying manually few hundred times. I wonder what "few hundred" means? 200 or 900?
I think I had a similar response rate when looking for work few months ago. I applied to ~40 jobs and I got 3 interviews.
I saw some YouTube video, where someone was applying to 500+ jobs per week with some automation in the same fashion. So this is probably already quite common. Makes the need for building ones own network even more important at this point, there's really no way to differentiate yourself from 10k people who apply to a job in the first 5 hours of its being posted, other than a internal reference.
93 comments
[ 386 ms ] story [ 3300 ms ] thread(my main empathy will go at the recruiters, having to filter out massive amounts of auto-generated applicant files!)
because it's an arms race, recruiters are also using AI to filter people, just look at the chat GPT hack for resumes
Doesn't even have to be anything big, just that there aren't enough keywords, or that there were gaps between employment that were too large, etc.
The ATS would then rank the applications and score them, and HR just looks at the top 15 or something and picks the top 5, whittles them down to 2-3.
They're not spending 8 hours a day reading resumes.
I guess it’s time to bring me to the farm
For those pigeons where there are no holes, it instead seems impossible.
I even applied to a place I worked a ~5 years ago, which I got via cold application. Apparently I’m not good enough anymore because I couldn’t get past the filters.
I finally found a job through a referral. Oddly, it was working with tech I have basically no experience in and probably had the most awkward interview out of all the ones I ended up doing.
I found a local contracting opportunity that sounded really interesting and something I could probably be a good fit for. It was a small company, 4 people, so I actually emailed the president directly with a well thought out introduction and explaining why I thought I would be a good fit for the position. Radio silence…
So, even the old adage of ‘hit the bricks and knock on doors’ has gone the way of the dodo.
I was legitimately considering going to their office but it being a small company, in the medical field, and uncertainties about how individuals might react with COVID still present…I opted for a personal email to the owner of the company. Maybe I should have opted for the ‘cold call’ approach and just went to their office but I erred on the side of caution/ decorum.
The actual hire will be someone the hiring manager knows from college.
And history has clearly proved that's just not the case.
Most resumes follow a template and are scanned for keywords and action-phrases. ATS software isn't smarter than that, outside of maybe making some guesses about you based on time worked at each job or if you're lacking attention to detail by not ending each bulletpoint with a period.
The time before that, I think I applied to three places and got two offers.
So the "normal" approach has been quite good for me for the last seven years. My success rate is orders of magnitude larger than 20/5000.
remote is here to stay, and AI will be eating a lot of jobs.
call my bluff: throw out 3 resumes this week and see what sticks. How many come with the same level of salary -- it's dropping, aggressively, across the market -- or more?
I'm well aware, but the article we're responding to is an anecdote as well.
> remote is here to stay
thankfully, it's been great for me
> call my bluff: throw out 3 resumes this week and see what sticks. How many come with the same level of salary -- it's dropping, aggressively, across the market -- or more?
I don't see the need to do this, I'm happy at my job and don't even know who you are. You're probably right, why would I waste my time?
"Hey, do you want to apply to the position? Tell me how many Vitae (tm) tokens you want to pay for it."
It doesn't need to be worth a lot, it just needs to be something.
The token itself has no monetary value. I'm literally meaning it as a way to be an utility token, which is just another way to say "funny money".
The idea is just to have a way to avoid spam applicants. There will be no point in hoarding it, unless you are just trying to stop everyone else from ever submitting job applications.
Sure. Less of them means more success for me. Or for my people, i.e. we here at Stanford hoard it in order to ensure the success of Stanford grads.
In principle I'm alright with the idea -- force people to limit themselves to 25 applications, stuff you really want, keep down the noise -- but what happens if people don't get a bite at 25? what are you forcing the hiring orgs to do?
No, less of them means that recruiters will not receive any applications in that channel. If recruiters stop caring about that channel, there is nothing stopping them from using or even creating a different one.
The micropayment problem is not new, and there's no indication that the Ethereum-based tokens have solved them, especially not with transaction fees that high. If you're talking about a non-decentralized solution, then as the sibling comment argued it boils down to an entry in a database and that's trivial anyway.
Use a testnet.
We are talking about an application with literal zero monetary value. No one will attack a test net to be able to mint themselves a token meant to be used solely for priority-rating in a process queue.
> $0.001 USD
We can't have it both ways: either paying a token is a disincentive to spam job postings, which means the token actually has monetary value that's higher than zero; or it's not, then it's useless.
> (either) the token actually has monetary value that's higher than zero; or it's not, then it's useless.
Are you familiar with Germany's Pfand system?
You go to a beer garden, get your pint which costs (say) 5€. The bartender asks you for 7€, meaning a 2€ "pfand". You get your beer and a physical, plastic token.
When you are done with your beer, you go back to the bar and if you return your glass with the token, you get your 2€ back. The key thing is that you need to bring both. Lost the token? No cash back for you, might as well take the glass home. Lost/broke the glass? The token is worthless.
The point here is that the token, by itself, has no value. It's sole purpose is to be a unit of accounting.
Similarly, having job tokens independent of any kind of deposit or incentive will not prevent anyone from spamming applications.
People will still have to find a way to get the tokens. They are not going to be creating them out of thin air.
> It doesn't need to be worth a lot, it just needs to be something.
Market pricing, though. I can see the rate for being considered for lucrative tech jobs climbing into tens of thousands of dollars (or one or two paycheques).
"The tortoise lies there, its belly baking in the hot sun, beating its legs, trying to turn itself over, but it cannot do so without your help. You are not helping. Why?"
Thanks I hate it.
This is seems to be a good example of AI enriching/assisting with other processes.
Technology makes it easier to be more anonymous, superficial, and lazy on both sides of the equation. Each side will gain a small temporary advantage with every advance in AI and the end result will be that EVERYONE will be worse off than before.
That's because the job application process will be worse, people will feel more isolated, and you'll still need to spend just as much time as before if not more learning all these arcane tools.
Technology is truly becoming a disease. We've taken something good and used too much of it like sugar. Everything has a happy medium, and technology used with moderation could have been something good, but its use is being pushed to its logical conclusion through these prisoner's dilemma-type situations and is resulting in a degradation of society.
The people that make AI make me sick.
Without AI, such a process favors potential employees that are good at writing likely word sequences themselves, and not those that are good at actually dpoing the job.
That's how most companies do it, even today. There is a special subset of companies that overautomate the hiring process, but it's not most of them (outside of SV-style companies, anyway).
It’s as if suddenly a significant number of ai people are hell bent on spam, ip theft, putting people out of jobs, spreading fear, manipulating and gaslighting folks, and so on. Almost as if after decades of irrelevance and frustration at memes about if then else “ai” attempts they seek revenge.
The tone was set by their thought leaders, in particular sam altman and the openai clique.
Taking a tech that was supposed to be a force for good and turning it into a toxic charade is quite the achievement. But gradually people are starting to push back.
People like Sam Altman just go after short-term profit because they can. They are psychopathic.
It seems to me like a good example of AI making something worse for everyone involved.
and I don't care if large corps with automated ATS systems are mildly inconvenienced.
Statistically, most people weren't getting hired anyway with standard, non-AI ATS systems + the Mk1 Human Eyeball.
And of the hires, if you define successful hire as 2+ years in the role, most new hires fail at a rate between 50-80%[1]. Clearly not having AI is a failure. It might create tons of spam, but there is already a ton of spam, and it's miserable for everyone.
[1] https://itworkforce.intersog.com/blog/why-do-50-percent-of-a...
Neither do I. However, this use of AI harms more than just them -- it harms all other job-seekers as well. It looks to me like doing this just accelerates the race to the bottom for all.
A legit use of the power of AI might be to scan 5000 listings and present you with the 20 best fits.
It wasn't even about AI at all except as example or set dressing. You can substitute "computer" or "automation" for "AI" in my comment.
The point was about "value" that is actually just external cost imposed on others.
Mind you, I've never applied for a job myself, it's all gone through recruiters. This shows that that's going to continue, it's real people knowing real people, networks etc.
Instead of focusing on quantity - number of applications - I decided to focus on quality - better resumes that emphasise the candidates skills that align with the job ad.
And instead of applying to all job ads, my service analyses a candidates resume and finds positions that closely match the candidate experience.
I think I had a similar response rate when looking for work few months ago. I applied to ~40 jobs and I got 3 interviews.