Member-only story
Can ChatGPT-3 fix hiring?
I can see GPT-3 optimizing the screening and hiring of contractors and technical employees. Here’s how.
Screening and hiring candidates is usually a suboptimal process.
At best, you end up solving the secretary problem. Even so, the best solution to the secretary problem only finds the best candidate for the job just 37% of the time.
Hiring is suboptimal for a few mutually reinforcing reasons.
Why is hiring suboptimal?
In person screenings
For one, screenings have to be done in person.
You could use l33tcode tests and other automated processes but ultimately, someone has to get on a video call with a handful of candidates to interview them in person. Every candidate who is a poor fit for a role that someone has to screen in person costs money.
Variability in judgement
The second problem with screening is that not everyone is judged the same way. There is too much room for unevenness in how interviews are conducted, even by the same interviewer.
For example, an interviewer may end up phrasing the same discussion points in different ways to different candidates allowing one candidate to outperform another purely on the basis of chance.
An adjacent point is that different interviewers screening on the basis of the same discussion points may interpret similar candidate responses in different ways. There is no easy way to gauge how evenly screeners apply the same metrics and measures across multiple candidates so that each candidate is judged the same way.
Screener’s Limited Knowledge of the problem space
The third issue is that an HR screener is not always suitably equipped to gauge a candidate across multiple dimensions. At best, an HR screener can run through a checklist of valuable jobs skills and forward a candidate to the next round. This ties back to an earlier point that every poor candidate forwarded up the interview chain costs money and emotional strain to the interviewing team who start despairing at candidate quality.
