These are truly maddening. I’ve recently fallen prey to another trick - email from a famous-ish tech person starting a new company and asking to chat, but then responding with interest just enters you into the normal pipeline with a hiring manager. Nothing wrong with that pipeline, but don’t misrepresent the outreach as an opportunity to chat with an influential person when it’s actually just an automated recruitment email.
It can be worse: Someone recently posted about going for an interview and discovering that an AI was interviewing them, complete with uncanny valley imitation human over the video call. This was disclosed to candidate, along with the strict injunction that if the candidate used AI in any way for the interview, the process would be terminated immediately.
"Impersonal lack of human touch for we, but not for thee."
This is one of my bigger issues with generative AI.
It used to be easy to spot effort: a cool logo on a business card, an interesting website design, a recruiter email that speaks to a specific experience I have on my LinkedIn profile. This used to indicate that someone was willing to do a bit more than the bare minimum and gave them some legitimacy.
Now I see any of these things and I have to do the extra work to try to figure out if it was AI generated.
You can still easily spot actual work, whether in website design or recruiter messages. Isn't that the whole schtick of the article?
But low-effort spam will always dominate, that's just the nature of it. You will always receive 10 times more messages that were only 1/10th of the work on the other side. How could it ever be otherwise?
That’s what makes this so annoying. I don’t mind a template, but the fake personalization gives it a whiff of deception in a way a pre-LLM simple template mass mailing doesn’t.
Sharing a trick I learned here on HN: I prefix my first name on LinkedIn with an emoji (green check box).
Whenever a message starts with “Hi (emoji) (first name)” it’s 100% spam.
So far they are not training to fix this (all they have to do is drop the prefix), which is somewhat surprising. Zero messages that address me correctly seem automated. Maybe because in today’s day people can legitimately have emoji in their names?
I keep getting this one that's like "Hi, our Chief Engineer @ <Massive DefenseTech startup> has personally reviewed your resume and wants to speak with you ASAP".
Never mind that I've never sent them my resume, and my LinkedIn is bare bones with only a job history and no description or skills. I investigate further and the Chief Engineer of this program is in his mid-20s and has only ever worked at said DefenseTech startup.
Also, keep a curated list of headhunters that you respect.
I once was talking to a headhunter about a position he had (IIRC). I wasn't interested, but mentioned someone else he should look at. When he found out where my contact was currently working, he said, "Oh, they're a client. I can't recruit from there."
He went on my list. A headhunter with ethics about not stealing employees from his own customers? That's good enough to make him someone I want to talk to when I'm in the market.
Keep adding to the list on the (rare) occasion that a headhunter does something that catches your notice. Takes five minutes a year, if you're lucky. When you're next looking for a job, it can save you days.
there are good recruiters out there; if you find one they'll probably get you several jobs over your career. The baseline is so shockingly low that you'd think more recruiters would put in the exceeding low level of required effort that would put them in the top 5%. Go figure...
I actively report these as spam (they are) and block the email. +1 on insulting intelligence.
> they probably haven't spent more than 10 seconds looking at my profile to see if it’s a fit.
I doubt they even glance at your profile, probably scraped and stored somewhere that's sold to these AI recruiting / sourcing companies the recruiters use.
I assume it’s like how scams have typos to self-select for people who are not paying attention. Presumably any candidate who cares is not worth going after
I get tons of these. I suspect they must run on crappy local models to save money, because the ones I get are also terrible at drawing a simple line connecting my experience with the job.
I look at them with a mixture of pity and contempt. They are examples of "In the land of the blind, the one-eyed man is King" work: i conjecture that this type of "shallow customization" is going to be most persuasive with recipients who have very little exposure to slop and/or the kinds of things human recruiters used to say when visually scanning some stale copy of a résumé while on the phone with the prospect.
I get ones for projects specific languages that are not on my resume. Never mind crappy models, they could rule me out with traditional keyword matching.
Like robocars killing people, it feels worse when it's AI.
I mean, before it was humans using template text with dynamic fields; in some cases the first time the sending human would actually see the message would be if you responded.
But now you can get all the way to the first call booking without a human ever being troubled to participate in the interaction.
In my experience, LLMs have not improved the quality of targeting (let's assume for each person doing outreach there are at least a few people who actually want the messages being sent), and my inbox is absolutely flooded with AI generated campaigns. I have stopped unsubscribing and I just mash the spam flag.
These are truly maddening. I’ve recently fallen prey to another trick - email from a famous-ish tech person starting a new company and asking to chat, but then responding with interest just enters you into the normal pipeline with a hiring manager. Nothing wrong with that pipeline, but don’t misrepresent the outreach as an opportunity to chat with an influential person when it’s actually just an automated recruitment email.
It can be worse: Someone recently posted about going for an interview and discovering that an AI was interviewing them, complete with uncanny valley imitation human over the video call. This was disclosed to candidate, along with the strict injunction that if the candidate used AI in any way for the interview, the process would be terminated immediately.
"Impersonal lack of human touch for we, but not for thee."
Even better is when you get put in touch with the person's OpenClaw (or w/e agent) writing on behalf of them, as them.
Everything is wrong with this pipeline for the reason you noted. It is unacceptable.
This is one of my bigger issues with generative AI.
It used to be easy to spot effort: a cool logo on a business card, an interesting website design, a recruiter email that speaks to a specific experience I have on my LinkedIn profile. This used to indicate that someone was willing to do a bit more than the bare minimum and gave them some legitimacy.
Now I see any of these things and I have to do the extra work to try to figure out if it was AI generated.
You can still easily spot actual work, whether in website design or recruiter messages. Isn't that the whole schtick of the article?
But low-effort spam will always dominate, that's just the nature of it. You will always receive 10 times more messages that were only 1/10th of the work on the other side. How could it ever be otherwise?
That’s what makes this so annoying. I don’t mind a template, but the fake personalization gives it a whiff of deception in a way a pre-LLM simple template mass mailing doesn’t.
Sharing a trick I learned here on HN: I prefix my first name on LinkedIn with an emoji (green check box).
Whenever a message starts with “Hi (emoji) (first name)” it’s 100% spam.
So far they are not training to fix this (all they have to do is drop the prefix), which is somewhat surprising. Zero messages that address me correctly seem automated. Maybe because in today’s day people can legitimately have emoji in their names?
Emoji on linkedin is itself a tell of llms
I keep getting this one that's like "Hi, our Chief Engineer @ <Massive DefenseTech startup> has personally reviewed your resume and wants to speak with you ASAP".
Never mind that I've never sent them my resume, and my LinkedIn is bare bones with only a job history and no description or skills. I investigate further and the Chief Engineer of this program is in his mid-20s and has only ever worked at said DefenseTech startup.
We need to have RCAPTCHA where the recruiter is quizzed on job images.
CLICK ON ALL IMAGES OF DEVOPS PIPELINES
CLICK ON ALL THE OKR IMAGES
And block on pass, because real recruiters wouldn't know any of this, so correct answer indicates it's a bot.
Invest in your actual network. It makes ignoring recruiters extremely easy. I try to remind myself of this.
Also, keep a curated list of headhunters that you respect.
I once was talking to a headhunter about a position he had (IIRC). I wasn't interested, but mentioned someone else he should look at. When he found out where my contact was currently working, he said, "Oh, they're a client. I can't recruit from there."
He went on my list. A headhunter with ethics about not stealing employees from his own customers? That's good enough to make him someone I want to talk to when I'm in the market.
Keep adding to the list on the (rare) occasion that a headhunter does something that catches your notice. Takes five minutes a year, if you're lucky. When you're next looking for a job, it can save you days.
there are good recruiters out there; if you find one they'll probably get you several jobs over your career. The baseline is so shockingly low that you'd think more recruiters would put in the exceeding low level of required effort that would put them in the top 5%. Go figure...
I actively report these as spam (they are) and block the email. +1 on insulting intelligence.
> they probably haven't spent more than 10 seconds looking at my profile to see if it’s a fit.
I doubt they even glance at your profile, probably scraped and stored somewhere that's sold to these AI recruiting / sourcing companies the recruiters use.
My solution to this spam is that I changed one letter in my LinkedIn name to a Unicode that looks like a letter but isn’t (think math symbol, etc).
If I get email or a message that includes that character, it proves software sent the message and it wasn’t typed by a human.
you're using the domain-typo squatting trick for good - kudos!
Won't that also block anyone who copy-pastes your name? I often copy paste names so that I don't make typos.
Yeah, even if the name is very familiar, botching that first-impression makes it relatively high-stakes.
It only gets worse from there when I'm trying to contact Krzysztof Wiśniewski.
Is it more impersonal than all the ones I still get with "Hello [insert name]...".
I assume it’s like how scams have typos to self-select for people who are not paying attention. Presumably any candidate who cares is not worth going after
I get tons of these. I suspect they must run on crappy local models to save money, because the ones I get are also terrible at drawing a simple line connecting my experience with the job.
I look at them with a mixture of pity and contempt. They are examples of "In the land of the blind, the one-eyed man is King" work: i conjecture that this type of "shallow customization" is going to be most persuasive with recipients who have very little exposure to slop and/or the kinds of things human recruiters used to say when visually scanning some stale copy of a résumé while on the phone with the prospect.
I get ones for projects specific languages that are not on my resume. Never mind crappy models, they could rule me out with traditional keyword matching.
y'all are getting recruited?
Like robocars killing people, it feels worse when it's AI.
I mean, before it was humans using template text with dynamic fields; in some cases the first time the sending human would actually see the message would be if you responded.
But now you can get all the way to the first call booking without a human ever being troubled to participate in the interaction.
In my experience, LLMs have not improved the quality of targeting (let's assume for each person doing outreach there are at least a few people who actually want the messages being sent), and my inbox is absolutely flooded with AI generated campaigns. I have stopped unsubscribing and I just mash the spam flag.
> That's the gap.
That's one of the shortest LLM tells I've seen.