> So why should I keep doing what I'm doing when it's getting me nowhere? Why shouldn't I switch to an automated "shotgun" approach that applies me to as many jobs as possible to which I vaguely fit the requirements?
I’m in a big semi-private Slack where people have been discussing CS application strategies for a long time (since before ChatGPT).
The desperate people usually go through an arc where they try automated applications and embracing LLMs. Their response rate is dismal, but they make up for it with shotgun volume.
The catch is that when they finally get a job, it’s usually at a company that sucks. Some place with incompetent hiring managers who can’t tell the difference between LLM slop and a genuine application. Interview processes that leave so much room for LLM cheating that all of your coworkers are going to be LLM jockeys too.
So you can try it. You might get something out of it, which is better than nothing. However, if you’re expecting a good job at a good company then it’s not going to deliver what you expect.
This is just the first pass. There are second pass strategies that could improve and are even more insidious:
- review your generated CV pre-submission, make changes, do this a lot. Eventually you'll have a training set to fine-tune the model
- throw 100-200 CVs at a job and see what sticks. That's your training set for that job. Now you have tuned the hiring manager's preferences. Follow up with your actual CV. Side benefit is it will jam up other candidates.
This is just fear mongering. If a job posting got spammed with 200 fake resumes from multiple fake applicants then the first thing we’re doing is cancelling our job postings with whatever service is so poor that it can’t reject basic spam attacks like this.
Honestly, I think people vastly overestimate how much hiring managers use AI for filtering. Blaming AI for rejections has become a common coping mechanism because it’s easier to think that a broken AI filter rejected you instead of the company making a valid decision to go with someone else.
> throw 100-200 CVs at a job and see what sticks
If your experience wasn’t good enough the first 10 times, doing another couple hundred rounds of LLM word manipulation isn’t going to make it better.
You don’t need to blame “AI” (or LLMs specifically) for the rejection mess that, good old fashioned ATS (applicant tracking systems) already automated rejection either outright or due to selection priority, filtering for keywords or phrases, biasing towards certain more easily parseable document formats, and so on was already happening around 2018-2019, probably before.
And resume refinement representing and reformatting essentially the same information has always been a commonplace trick to improve your odds. My simple first pass resumes around that time must have never seen the light of human eyes because optimizing things around such systems, adjusting formatting, pushing docx versions, and so on increased my return response rate per submission for the exact same information. People just tend to forget they’ve gone through such processes or are moving positions through networking. The cold market has been abysmal for quite some time, even if you’re qualified.
Naysayers haven’t been submitting to cold options I suspect which is why the trend has always been denial. But with mass layoffs, people are having to resort to cold application processes and finally experiencing at scale how terrible the process has been.
> Naysayers haven’t been submitting to cold options I suspect which is why the trend has always been denial. But with mass layoffs, people are having to resort to cold application processes and finally experiencing at scale how terrible the process has been.
Aye, this. Got all of my jobs historically though word of mouth. Sat next to someone at a wedding reception, or talks at a Linux User Group, or colleagues from one job going to the next and pulling everyone with them, etc.
cold applying was brutal. worked out, eventually, but it feels/felt like such a waste of time.
see comment below- "
belinder 1 day ago | parent | prev | next [–]
I was hiring manager for 3 positions about 4 months ago and the amount of fake applications out there was mind boggling to me. I would say 90% were either entirely fake or had the exact same generated ai text. It got so bad that we started only looking at resumes that had a working LinkedIn link.
Also after so many bad resumes I started being very forgiving for the ones that didn't fully match the job requirements if they had something in them that made it seem like a real person, e.g. a personal hobby section. I think a lot of people discourage writing that but I argue it makes you stand out in an ocean of fake and copy pasted junk."
Hiring managers don’t have infinite time and resources, they’ll just pursue other more fruitful avenues where a DoS attack isn’t possible.
This is a great way to entrench the recruiter middleman further though, because paying them a 20% cut to bypass the bullshit is already what they sell (and sometimes deliver).
Unless the place has had 100% turnover in the last two years it sounds a bit dubious. Even some of the worst places to work that I know of haven’t churned through their entire development staff since ChatGPT first released.
I’m in a big semi-private Slack where people have been discussing CS application strategies for a long time (since before ChatGPT).
The desperate people usually go through an arc where they try automated applications and embracing LLMs. Their response rate is dismal, but they make up for it with shotgun volume.
The catch is that when they finally get a job, it’s usually at a company that sucks. Some place with incompetent hiring managers who can’t tell the difference between LLM slop and a genuine application. Interview processes that leave so much room for LLM cheating that all of your coworkers are going to be LLM jockeys too.
So you can try it. You might get something out of it, which is better than nothing. However, if you’re expecting a good job at a good company then it’s not going to deliver what you expect.