Written to Attract, Read as Instructions

2 min read

AI screening is the first reader to take the job description literally.

Job descriptions have always had two readers: candidates deciding whether to apply, and internal teams deciding what they're hiring for. AI screening adds a third, and it's the only one that takes the document literally.

Recruiters screening by phone rarely work from the JD. They carry the working brief in their heads, built from past placements and pattern-matching, fill the JD's gaps without noticing, and adjust when an answer exposes the real requirement. The written version can stay vague because nobody relies on it.

An AI interviewer relies on nothing else. Vendors now compete on follow-up questions, probing an answer rather than ticking it off. Probing needs a standard: what a strong answer sounds like, where adequate ends, which trade-offs the hiring managers will accept. Most job descriptions don't contain that, and most intake meetings don't produce it.

The firms reporting results from these tools show this in the detail of their case studies. The criteria were codified before the agent made a call, and someone is paid to keep them current. That's role definition, relocated rather than removed.

The failure mode changes too. Consistency is the selling point, but a consistent screen scored against prerequisites rewards candidates who match the list. High performers rarely present that way. Their case sits in what they'd change, and it only surfaces in a conversation that follows them somewhere the criteria didn't anticipate. An agent screens them out the same way every time and leaves transcripts that look rigorous.

Director level is where the gap costs most. The criteria that decide success there, such as scope, judgement under ambiguity, and how people earn credibility with peers, are the ones least likely to be written down.

Automation doesn't remove the need to define the role. It removes the people who were quietly doing it.