Photo via greenhouse.com
Resume screening may wrongly filter out 98% of job applicants today, Daniel Chait, co-founder and CEO of Greenhouse, told Auren Hoffman on Summation.
Most of the argument about AI in hiring is over whether it will judge candidates unfairly. Daniel's point is that both sides already use AI against each other, candidates to send applications in bulk and companies to filter them, and that the fix is using it to give every applicant a real interview.
"The current state of hiring is so bad and so broken that the possibility to be less bad than that seems like a no-brainer."
Daniel coined the phrase "doom loop" for that cycle, and he has been collecting data on interviews since he hired engineers on Wall Street with a tape recorder in the room.
The full interview is covered here so you can skip it. 1 hour of audio, 9 minutes of reading.
Here are the 8 lessons that matter.
Summation with Auren Hoffman · Oct 27, 2026 · 1h · 51m saved
With Daniel Chait, Co-Founder and CEO of Greenhouse
Key Takeaways
AI on both sides of hiring has created a spam cycle, leaving "the vast majority of applicants unread"
Resume screening may wrongly cut 98% of candidates, and Daniel thinks a voice AI interview for every applicant could bring that toward 20%
The best test of an AI-native candidate is what they built in the last two months, not what the resume lists
Questions about past behavior predict performance best, and trick tests such as the iced-coffee interview predict nothing
AI displacement is real in entry-level legal work but slower than expected even in outsourced back-office jobs
"Never give up" is bad founder advice, Daniel says, and venture capital is paid to keep repeating it
1. The AI Doom Loop
Daniel laid out the cycle he named. Candidates use AI to send applications in volume because the process feels hopeless. Companies respond by using AI to cut thousands of resumes to the 25 or 30 a human will see, and that filtering pushes candidates to apply even more widely.
Daniel blames the first generation of AI hiring tools, not AI itself, for the doom loop
"The problem is not AI... I think there's a lot we can talk about the power of AI, but also there are risks and downsides to using AI, and I think we're seeing that in the job market today." — Daniel Chait
Both sides want the same thing, faster and more accurate signal, so the fix has to work for candidates and companies at once. Some filtering is legitimate, as one manufacturer's experience shows.
"I literally had a conversation with somebody that owns a pet food manufacturing business in Wisconsin, and they were getting applicants to work in the pet food factory... who lived in Southeast Asia." — Daniel Chait
