3 ms·

You sound like you think you can tell the difference between human-written and LLM-written resumes. Spoiler: you can't. So you've put up yet another arbitrary h

by feoren 2y ago

You sound like you think you can tell the difference between human-written and LLM-written resumes. Spoiler: you can't. So you've put up yet another arbitrary hoop to hiring for no reason, just like everyone else. Getting hired at your company is a game of getting through your weird roadblocks, with little to do with being qualified. Just like at every other company.

skybrian 2y ago
This sounds more like low-effort spam filtering than a hoop that’s likely to block legit applications.
- kkarakk 2y ago
  the moment i heard chatgpt can make resumes i knew no one was gonna bother with latex ever again. companies are gonna have to adapt to this - literally everyone on the planet is getting bombarded with "use AI or perish" messaging, why would they not use llms to format/create/write resumes?
  - skybrian 2y ago
    And I think it’s fine, but what they’re filtering on with their ChatGPT canary is people who don’t proofread the output to make sure it’s a resume they stand behind. It should be easy since it’s all about you, right? When using AI, “distrust and verify” is the best policy.
floobertoober 2y ago
They're saying they put a poison pill in the job listing itself, so if an applicant copy and pasted the entire job listing into an LLM the generated text will have some kind of "tell".
sebastiennight 2y ago
We didn't add any hoop, we just designed a honeypot. Other people in this thread have gotten it right. This is the equivalent of a hidden HTML field which spambots would fill, but humans wouldn't see. The rationale is not that we don't want any LLM users to be hired (which would be weird for an AI company). The rationale is that we don't want people who use ChatGPT to do their homework for them, without even reading the result of that. Especially since ChatGPT will happily lie in a cover letter to make your "experience" match any job posting, which is a form of fraud. So we "poison" the job ad in a way that normal humans will simply skip over, but LLMs will clearly have a ridiculous "tell" (which you would catch if you did proofread the result).

by feoren 2y ago

skybrian 2y ago

This sounds more like low-effort spam filtering than a hoop that’s likely to block legit applications.

kkarakk 2y ago
the moment i heard chatgpt can make resumes i knew no one was gonna bother with latex ever again. companies are gonna have to adapt to this - literally everyone on the planet is getting bombarded with "use AI or perish" messaging, why would they not use llms to format/create/write resumes?
- skybrian 2y ago
  And I think it’s fine, but what they’re filtering on with their ChatGPT canary is people who don’t proofread the output to make sure it’s a resume they stand behind. It should be easy since it’s all about you, right? When using AI, “distrust and verify” is the best policy.

floobertoober 2y ago

They're saying they put a poison pill in the job listing itself, so if an applicant copy and pasted the entire job listing into an LLM the generated text will have some kind of "tell".

sebastiennight 2y ago

We didn't add any hoop, we just designed a honeypot. Other people in this thread have gotten it right. This is the equivalent of a hidden HTML field which spambots would fill, but humans wouldn't see. The rationale is not that we don't want any LLM users to be hired (which would be weird for an AI company). The rationale is that we don't want people who use ChatGPT to do their homework for them, without even reading the result of that. Especially since ChatGPT will happily lie in a cover letter to make your "experience" match any job posting, which is a form of fraud. So we "poison" the job ad in a way that normal humans will simply skip over, but LLMs will clearly have a ridiculous "tell" (which you would catch if you did proofread the result).