I was interviewing for a job as a customer service agent with Anna. She had a low, pleasant voice and she’d nailed the pronunciation of my name—something few people do. I wanted to make a good impression except I had no idea what Anna was thinking because Anna couldn’t think. Anna wasn’t technically a person. She was AI.

Not only is AI changing how we do our jobs, it’s also changing how we get jobs. This ranges from using AI to screen resumes, schedule interviews, even conduct them. According to a 2025 report, 20% of companies are using AI to interview candidates.

Even so, nothing can replace human recruiters, the folks who’ve deployed Anna into the wild stressed to me. After I spoke with her, I quickly understood why.

AI Anna clocks in

Even though I wasn’t really interviewing for a job—this was all an exercise for this story, of course—I was still nervous.

I asked the team behind Anna to provide a job description so I could prepare, but outside of this experiment, I was sadly lacking in actual customer service experience. I also didn’t know how AI Anna was going to react to awkward silences, panicked misdirection, or if she’d be able to tell if I was lying. These worries are bad enough with a human. How would a computer program react?

I got on the phone and connected with Anna. She was pleasant, and frankly, sounded way more human than I was expecting. We exchanged greetings, and before long, I was in full-on job interview mode with an AI.

First up, she asked me to describe a time when I had to explain something complex over the phone clearly. I blanked. Finally, I described how journalism involves explaining complex ideas because you’re asking questions. It sounded weak even to my own ears.

Sure enough, she was not impressed. “I’d like to explore a scenario that’s more specific to the role we’re discussing,” she replied firmly.

Fair point. I managed to dredge something up from a high school job. Mercifully, AI Anna accepted the answer and moved on.

Next, AI Anna wanted me to talk about a time when I had to problem solve for a customer. This, I could answer. I dove into my brief stint organizing a literary conference where writers paid to meet with agents. Occasionally agents went astray because they were hungover or running marathons and I’d be left to find alternatives like rescheduling—

Anna cut me off.

“That sounds like a high-pressure situation. . . . It’s great that you were able to come up with alternatives. Now I’d like to switch topics for a moment.”

Yikes. I wasn’t ready to switch topics, but AI Anna was, and I couldn’t tell why. Was my example off topic? Was I taking too long to answer her question? Before I could ask, Anna had already swept on to background checks.

I invented a criminal background and told AI Anna I had done some time in prison. She thanked me for being honest, and told me that she could not make any decisions. She said candidates with a criminal record would be considered on a case-by-case basis (something that would have to be verified by a human).

Then I wanted to know if I’d be required to work overtime. She let me know I’d be required to do overtime the first six months, but only one or two times a month. Needed, accurate information that couldn’t just be googled—great.

Honestly? While I found her transitions a bit jarring sometimes, she handled most questions with aplomb.

How we got here

AI Anna is the product of PSG Global Solutions, a staffing firm. Before deploying AI Anna in the market, the firm asked Brian Jabarian, a researcher at the University of Chicago Booth School of Business with doctorates in economics and philosophy, to study the AI Anna’s effectiveness. (Jabarian received no funding from PSG).

In a study released in September, Jabarian conducted an experiment where 70,000 applicants for a customer service job were randomly assigned a human interviewer, an interview with AI Anna, or the ability to choose between the two.

The results are surprising, and surprisingly promising for the candidates.

AI interviews resulted in a 12% increase in job offers, and a 17% increase in 30-day retention on the job. Moreover, when offered a choice, 78% of applicants chose to be interviewed by AI. Jabarian theorized this was because the AI was easier to schedule with: job applicants who needed a job quickly could book a call immediately.

Why the positive outcome data? Jabarian pointed out that, on average, an AI interviewer got through more required topics than human recruiters since they couldn’t be distracted. (I mean, Anna did move at a brisk clip.)

“AI leads to a more consistent interview experience,” he said. “It lets the candidate talk more, and has a 50% chance of covering 10 of 14 required topics compared to 25% for human interviewers.”

AI Anna clocks out

Afterwards, I debriefed the interview with David Koch, PSG’s chief transformation and innovation officer.

First, he showed me AI Anna’s backend: The platform had generated a recording of our conversation, a transcript, a summary of the call (including suggestions for next steps, like a follow up to discuss my criminal background), and an overall recommendation: AI Anna thought I was qualified (yay!) but merited human follow-up because of my criminal background.

AI Anna also recommended a follow-up because she’d cut me off when I was talking about the literary conference. Koch explained my speaking cadence is a touch slower than average, and AI Anna is programmed to respond after a certain amount of time or else the flow of conversation can become jerky.

Koch noted that AI interviewing was better suited for some situations and not others. He recommended AI interviewing for high-volume hiring where there’s a need to source candidates quickly for jobs that are seasonal and high turnover, like customer service agents or travel nurses. Koch also said AI interviewing is best suited for cases where there’s less complexity, in which you don’t need to sell a candidate on a role.

From my standpoint as a lay person, the technology behind AI Anna struck me as marvelous. She corralled me into staying on topic, and was capable of social niceties. She provided detailed answers to all my questions.

For recruiters, this could be life changing. It’s not that AI Anna might replace them, per se (there were already things from the interview that a human colleague would have to address or follow up on). But recruiters could farm out tasks like screening calls to AI while they worked on more high-level tasks.

However, this made me worry. If AI Anna existed to save companies time, what happens to candidates who get flagged for follow-up, even for something as simple as speaking slowly—let alone a criminal background? If there are more than enough qualified candidates to fill roles, I imagine a harried hiring manager would make offers to people who don’t require follow-up. Exception cases that require more time, like me, might fall to the wayside.

The future: cold, but competent

After my conversation with AI Anna, I felt hollow. Typically, if an interview goes well, I have the high of having connected with someone who might make me feel valued, desired, and possibly in the mix for a new job. If it doesn’t go well, I spend the next couple of days wallowing in self-pity and dissecting potential red flags.

AI Anna’s preprogrammed human-like intonation left me nothing to go on. Did she like me? Or was meh on me, but still think I was qualified?

I couldn’t tell probably because AI Anna does not have emotions and did not care about me.

But how much does this matter? A Gallup study found that 44% of respondents said their interviews drove them to accept an offer or not. Ideally, candidates would be able to interview with their direct supervisor  before getting a job in order to suss out personality match—but for a screening interview, AI Anna’s value was undeniable.

She raised the floor for interview quality: She’s personable and she offers a consistent experience. There’s no need to worry about the mysterious intangible of “chemistry.” Jabarian also pointed out that AI interviews reduce gender discrimination by half.

Done right, AI interviewers could reduce bias and help qualified job candidates who may not perform well during interviews because they lack intangibles such as charisma.

Still, I missed talking to a human.