3 ms·

> How do you know if the CV is accurate or not? What does this have to do with AI? People have lied on their resumes long before ChatGPT existed.

by jmye 3y ago

colesantiago 3y ago
Well it has everything to do with AI, as this will make the application process completely worse, fake CVs and cover letters produced at scale faster than before with little effort which does not help the sifting process for hiring managers. You might as well (as I suggested before) not rely on CV / cover letters as a signal at all, and look at other means such as real world projects (with working links) that people use.
- nradov 3y ago
  That's just silly. Creation and submission of fake CVs was never an obstacle before. It was always super easy, and AI changes nothing significant. The reason most job applicants are fairly honest is that any major lies are likely to be discovered during the interview and background check process.
  - colesantiago 3y ago
    The point is this can be done at scale, and this is a real problem for companies that have lots of 'candidates'. So my argument is that CVs is no longer a reliable indicator anymore.
    - nradov 3y ago
      That is not a valid point. Candidates have always been able to submit fake CVs to many companies. Most of them don't do so, nor will they in the future.
      - colesantiago 3y ago
        > That is not a valid point. Scale and ease of effort is a valid and relevant point as it is easier than before for cover letters and CVs to be generated they are as good as dead for hiring managers. To solve this, do not even consider the CV to begin with and consider looking at candidate's real world projects, bounties and paid PRs. like for example tinycorp [1] or comma.ai [2] [2] Is similar to a take home except the challenge is public and the deliverable is evaluated on an error score which lower is better. Solutions will be different to other candidates and the best ones will put in the effort and have a lower error score, it is then you can decide to spend your time interviewing them. I would even argue that this is even more effective than an ATS that random candidates are sending their fake CVs and cover letters through. [1] https://tinygrad.org/ https://tinygrad.org/ [2] https://github.com/commaai/calib_challenge https://github.com/commaai/calib_challenge
- jmye 3y ago
  I still have no idea why you think this is an AI problem. What is the value in creating a fake resume via AI that doesn’t exist without AI?

> How do you know if the CV is accurate or not? What does this have to do with AI? People have lied on their resumes long before ChatGPT existed.

by jmye 3y ago

> How do you know if the CV is accurate or not? What does this have to do with AI? People have lied on their resumes long before ChatGPT existed.

colesantiago 3y ago

Well it has everything to do with AI, as this will make the application process completely worse, fake CVs and cover letters produced at scale faster than before with little effort which does not help the sifting process for hiring managers. You might as well (as I suggested before) not rely on CV / cover letters as a signal at all, and look at other means such as real world projects (with working links) that people use.

nradov 3y ago
That's just silly. Creation and submission of fake CVs was never an obstacle before. It was always super easy, and AI changes nothing significant. The reason most job applicants are fairly honest is that any major lies are likely to be discovered during the interview and background check process.
- colesantiago 3y ago
  The point is this can be done at scale, and this is a real problem for companies that have lots of 'candidates'. So my argument is that CVs is no longer a reliable indicator anymore.
  - nradov 3y ago
    That is not a valid point. Candidates have always been able to submit fake CVs to many companies. Most of them don't do so, nor will they in the future.
    - colesantiago 3y ago
      > That is not a valid point. Scale and ease of effort is a valid and relevant point as it is easier than before for cover letters and CVs to be generated they are as good as dead for hiring managers. To solve this, do not even consider the CV to begin with and consider looking at candidate's real world projects, bounties and paid PRs. like for example tinycorp [1] or comma.ai [2] [2] Is similar to a take home except the challenge is public and the deliverable is evaluated on an error score which lower is better. Solutions will be different to other candidates and the best ones will put in the effort and have a lower error score, it is then you can decide to spend your time interviewing them. I would even argue that this is even more effective than an ATS that random candidates are sending their fake CVs and cover letters through. [1] https://tinygrad.org/ https://tinygrad.org/ [2] https://github.com/commaai/calib_challenge https://github.com/commaai/calib_challenge
jmye 3y ago
I still have no idea why you think this is an AI problem. What is the value in creating a fake resume via AI that doesn’t exist without AI?