It’s a sad truth of the being a candidate today that applying to jobs is a volume game. Even if you are targeted, focused, reaching out to contacts and all the other LinkedIn-wall best practices, there are times when you just have to cold apply to any post that seems vaguely relevant and interesting. Add to this, many posts are hidden behind sometimes multiple layers of obfuscation - you are applying to the firm that has outsourced to a firm that may or may not actually pass your information down to a company where the hiring manager can ignore it.

The upshot is that sometimes you don’t know a lot about what you’re applying to.

So, this particular post looked interesting - it spoke to a lot of my skills but in a new area (clean energy) that I’ve had a lot of interest in but not any experience. It hit the salary, location, size and other criteria that worked for me. Then, midway through the standard rituals of job applications - the resume upload, the demographic questions, the visa questions, the cover letter field that no one fills out and no one reads - the prompt popped up. “Before proceeding click here and talk to our AI recruiting agent. When you are done come back and finish.”

In espionage, this is called a “honeypot.”

The posting was from an “AI Recruiting” firm. Its name is a portmanteau of a slang term for the act of masturbation as performed by men and a slang term for the act of masturbation as performed by women. Note to startups: get at least one teenage boy with a Reddit account to review your company name before it goes out to the world.

The name comes from their two AI agents - a “recruiter” and an “HR representative” -that act as the public interfaces of the company. The recruiter agent (Male Masturbatory Act) interviews candidates, collects their preferences and experiences and then through some opaque process passes their information on to the “in house” AI HR agent (Female Masturbatory Act) who connects them with the company. If this sounds like the plot of a William Gibson novel, that is because it is the plot of the William Gibson novel Neuromancer.

(There’s also something to unpack about the gender politics of having the outgoing candidate-facing recruiter personalized as male and the offstage in-house HR rep personalized as female.)

Once you’ve completed (yet another) profile and uploaded (yet another) resume, you are directed to chat with the recruiter agent. I was expecting a text chat - like one of the customer service chatbots on a website that you have to type “LET ME TALK TO A HUMAN” over and over again until you can connect to someone that can solve your problem. Instead, you have to do a 20-minute verbal conversation with the AI agent.

Now, I know that this conversation is probably going to train some new LLM that will inevitably used to screw someone and I’ll also probably be getting spam emails for the rest of my life, but I wanted to see how this process worked.

The agent’s voice had a vaguely Californian tech-bro affectation - like someone who doesn’t own a Cybertruck but has looked at least one with longing on the freeway. Beyond the voice and onanistic name, the company didn’t anthropomorphize the agent further. There were no Duolinguo-esque animations or LLM-generated faces. You speak to a gray ball on a screen.

The agent started with some basic questions, including how I found the site (I didn’t say “You sent me here from your bogus job post, asshole”), and about some biographical details from my LinkedIn profile. Then the preference questions started.

I saw a Stephen Fry interview once where he broke down a trick that British royals use to signal interest when talking to their subjects during a meet and greet by repeating the last line that the other person says.

“What are you doing next week?”

“I’m going on vacation.”

“On vacation! Excellent! To where?”

“To Peru.”

“To Peru! Wonderful! And what are you going to do there?”

etc.

The agent has learned this trick (or been trained on conversations that use it). It asked what I was looking for in a role (“I’m looking for senior technical leadership positions working on ambiguous problems.”) and would reply with something like:

“I can see you’re looking for senior leadership roles with big ambiguous problems. That’s fantastic!”

Is the agent trying to rizz me up? It was unfailingly positive about my experiences as I related them, always with the repetition followed by a compliment. It asked about more, almost what you would expect from a human recruiter interview, even asking for clarifying details from some of the projects I described.

Towards the end of the interview, the agent asked about my longer-term aspirations, with a tone more psychoanalyst than recruiter. This is when it started to … glitch. Until now the conversation had been pretty linear, with topic leading to topic all following the repeat, compliment, ask pattern. Once we got to the longer-term vision questions, it had problems absorbing the answers. It would ask me some question “Where do you see yourself in 3-5 years?” and after I answered it would loop back to a not identical but very similar one “That’s great, it seems to really align with your experience. Now, given your desire for ambiguity, what would be ideal for you in the next 3-5 years?”

This fugue state took a while - maybe the last quarter of the interview. Not quite long enough for me to close the browser tab in anger but definitely enough to wonder if the agent should do one of those tests where it has to label which picture is a giraffe.

Then it was done. The grey circle stopped pulsating and the agent posted a transcript of our conversation on to the page, and I was redirected to the original, questionable post. (And for what it’s worth, I never heard anything from them, so no idea whether it was real or not. The site has made a small handful of relevant job suggestions that I haven’t seen on other sources). Since then, I haven’t been dragged into any further conversations with the recruiter agent, so no more talking to a gray circle for me.

Is this type of experience the future of recruiting? It’s not hard to envision a dead-internet-theory type of interaction where candidate bots talk to recruiter bots who talk to AI bots. But (at least until all jobs are completely automated and no one does anything) these tools are ostensibly trying to, you know, hire a human. I haven’t seen much evidence they are any better at that than any of the other busted methods that are in play these days. Is there any evidence that candidates that get referred to the in-house agent are any better or have more success than other methods? None that I can see but I don’t have much visibility into that side as a candidate.

Like the “cultural fit” test I wrote about earlier, it seems like this is another attempt to scale the process of hiring, although this one seems more considered and personal.

Still, the company seriously needs to change its name.