Recent searches
Search options
not only are job candidates being put through AI filters now we are also starting to see AI phone screens: https://www.linkedin.com/feed/update/urn:li:activity:7302042746414845952/
this feels clumsy and inequitable
www.linkedin.comJulia Burke on LinkedIn: I recently submitted a resume for a great sounding position. A few minutes… | 228 commentsI recently submitted a resume for a great sounding position. A few minutes after submission, I receive a phone call. This call was from the agency placing this position. Cool, fine, great to have such fast response.
The issue is that it was an AI model calling me. And this AI could not understand what I was saying. It asked me very detailed questions such as: "Explain two times you had to create test plans for complicated systems and what that entailed" and yet trying to answer caused the AI model to run out of tokens, or freeze, or just hallucinate until it stopped functioning. I'm not entirely sure. I sat there in silence for a few minutes, just saying, 'hello? are you still there?'
I had no choice but to hang up, and call back. The model picked up again, knew who I was via my phone number, and picked back up only now it was going so fast through questions I had no time to answer, it would ask a question and then immediately respond, 'Got it! Great answer!' and move on. Then it was over, with a statement that if this AI model's notes and my resume were a match, a real person would contact me.
It's been a while. So clearly I wasn't a match even though I absolutely was, at least judging by the job description. The issue here is that I have no idea what information this AI model gathered, I don't know how it was tested (clearly not every much), I don't know what safety policies it has enabled, I don't even know if it wrote down anything to send off or was completely unresponsive in answer gathering. For all I know, it wrote down gibberish, or slurs.
I don't want to use LinkedIn as a diary or a place to gripe, I try my best to use this site solely for job searching and network outreach. However, I felt it prudent to share my experience because if we're up against AI phone screenings for positions, we're deeper in the job hunt hole than we realize.
I urge companies to properly test their AI models, and to be upfront about the AI process and give candidates some information about the AI model training. At the very least, implement a policy where the information that the AI has gathered, is emailed to the candidate so they can confirm the AI model is operating correctly and the candidate can submit edits back to the model.
Additionally, please understand the limits of your AI and pay attention to the questions asked. Asking me to go indepth about two times I created complicated test plans, in one question, is a huge ask and might not be the best question to ask, if the model is going to hit its constraints.
I test AI prompting and output, and AI cannot be left to just operate on its own especially as the first contact for your placement agency. So work with the candidates in return and we'll all be on better footing in this process and more than likely, your human recruiters will get more reliable candidate information back.
Also, if you are looking for someone to test your AI output, *raises hand*, I'm available, hit me up. | 228 comments on LinkedIn
@skinnylatte I’ve had to do a couple AI video conference interviews and they’re fucking awful
@shovemedia arghhhhh