Humans, AI Screening, and the Final Hiring Decisions 

Screening isn’t interviewing, but interviewing is still screening. 

Screening helps HR and talent acquisition know if the candidates have the skills and experience to be interviewed more in depth by the hiring managers (and maybe others). Employers across industries may vary how they approach screening, but that’s still the approach.

Interviewing is screening in and out further to find those the managers may want to make offers to and hire.

Over the years in our CandE Benchmark Research, we combined screening and interviewing into one stage for candidates to answer questions about. 

This year we separated screening and interviewing into two stages to see how different populations of candidates perceive their screening experience separate from interviewing, especially now that artificial intelligence is being used more to screen in and out. 

We’re still collecting data for the 2026 benchmark year, and it’s clear that salaried professionals, management, and senior leadership do not like AI phone screening (AI agent) or one-way recorded screening. It’s also clear that a very small percentage of salaried professionals, management, and senior leadership, as well as hourly, know they’re being screened by an AI agent. 

Do they know it’s AI screening them? Most likely they do, and for some, maybe not. But unlike when we ask candidates if they use AI tech to help them with their job search, it’s doubtful that most candidates would underreport being screened by AI. The negative sentiment in our benchmark research the past few years has been clear: rejected candidates believe that AI technology is screening them out. 

The big difference between the salaried job types and hourly candidates is that hourly rates an AI phone screen very high (81 NPS), while salaried professionals and management / senior leadership rate those screens 84%+ lower (20-33 NPS). Hourly is usually more transactional and any engagement is perceived positively. Also, the resentment rate for salaried and management professionals, the percentage of candidates who have a poor experience and state they’ll never engage those employers again, is 2-3 times higher than hourly (see table below).

But in this difficult professional candidate market, they aren’t having it. What’s AI is not human and therefore they can’t have reciprocal engagement. Even if the conversational AI is more sophisticated than binary questions and answers. 

The same is true for one-way recorded video screens, where many more candidates across job types have experienced them compared to AI phone screens. Our benchmark data has always shown that one-way recorded screens have been the least favorable for candidates. Even if it has an assessment component, there’s no reciprocal engagement for the candidates, and the NPS ratings are much lower. Hourly is still high at 45 NPS, but salaried and management are very low, at 19 and 0 NPS respectively. 

Overall resentment is high here as well, averaging over 20% across job types. That’s 1 in 5 candidates who won’t engage an employer again after the one-way recorded screen experience (and communication and feedback loop frustration). 

Does that mean we scrap these screening methods? No, not if they’re working for employers to screen in and out. The communication and feedback loop frustration referenced above could be mitigated by more timely and consistent communication and feedback, however. Period. 

Are you being transparent with candidates about why you’re using AI agents and/or one-way recording screening? 

If there’s an assessment component, are they getting any feedback from it?

If there’s not, are they still getting any feedback about their screens?

Are you emphasizing on your career sites and in your communications, that no matter the screening technologies used, humans are still making the final hiring decisions?

And are they?

This last question isn’t rhetorical; it’s literally at the existential core of what’s fundamentally changing about recruiting and hiring. 

For humans – HR, recruiters, hiring managers, and candidates – that’s not a good thing.