Hacker News new | ask | show | jobs
by mead5432 779 days ago
My partner has been searching for an SDE job for 8 months and recently received an email requesting her to do a recruiter screen. She responded and got a call instantly where the “recruiter was a text-to-speech AI.

Presumably the entire process was automated; I’m assuming if they’re using text-to-speech, the selection and initial email was all AI.

It’s an arms race on both sides.

6 comments

It's the curse of scale. Automation allows there to be plenty of job postings and job applications, but it sidelines the human part of actually talking to someone. Those in the know will end up sidestepping this altogether and using their networks to get a direct line to a real hiring manager or applicant. I've done this myself where I talk to past collaborators and professors I know and ask "Do you have an intern/soon-to-graduate-student you recommend with skills X, Y, and Z?"

This is admittedly unfair to outsiders, but if you talk to someone, there's a high reputational pressure that assures you're dealing with a real person who's likely a decent match for the job/applicant. This is valuable precisely because there's high friction and doesn't scale.

I generally agree that relying on a network and human referrals is the way to go but building an effective network is much harder than it used to be. Meetup groups and local conferences were a good source to make connections pre-pandemic but that has really withered in the last couple years.

Past collaborators are also a great source but she is junior and doesn’t really have any she can leverage.

My initial advice to her was do some real projects, document her work and insights in a blog and engage with others on social media but that’s mostly screaming into a void.

My approach to hiring is to focus on competencies (what do I need them to accomplish) rather than skills (how many years of JS) but that doesn’t cut down on the number of applicants any, just makes initial eval a little easier. I was pondering the idea of leaning in more on AI to screen on competencies which might incentivize higher information density on resumes but the quality of that eval might not actually be any better and only increase the “black box”-ness of it all.

My suggestion is to focus on a fewer number of companies, instead of sending resume in bulk. When you narrow your choices you can study well and notice bugs or issues that need to be addressed. Fix them, send the results to someone in the company, it will increase your chances substantially.
It's a sad day when the answer is "do unpaid labor for a chance to be noticed."
The people doing the interview are being paid but the company itself isn't making any money from the interview process either and it's dedicating resources that could be more productively used elsewhere for immediate results.

Both parties are making short term sacrifices in hopes of finding a mutually beneficial relationship.

When you have 500 candidates applying to the same job you have to find a way to stand out.
Tried this before, doesn't work.

I've had more success sharing financial models on printed paper, than on working on code for free, building plugins and the like.

> building an effective network is much harder than it used to be

If you’re in an American city, go to cafes and bars. Be the annoying person who asks what someone does and persist. It’s absolutely annoying. But I’ll also always follow up on it, because that skill in itself is differentiating.

Absolutely pathetic. Might as well become worm.
> Absolutely pathetic

It seems to be a class-based social more. Among the rich and upper-middle class, approaching someone to introduce yourself isn't unfavourable. If anything, it's seen as a right and children are taught to do it. Among the lower classes it's seen as uncouthe.

You see it strikingly at e.g. birthday parties and galas. (Particularly in the U.K., Western and Southern Europe. Though there, unfortunately, such cold introductions usually aren't enough to cross the barrier. Hence my qualification for this only working in American cities. Also Nordic countries. It even extends to the design of social spaces, with private clubs and even elite airline lounges having chairs face each other while tables at fast food places are more isolated.)

You rang?
That’s why we’re building Touvlo.co (for hardware engineers) - human engineer interviewers, interviewing humans on real world applicable engineering skills, more fair to outsiders, standardized, scaleable!
> It’s an arms race on both sides.

In my experience, the AI-vs-AI hiring battle takes place in parallel to normal hiring pipelines. Even before ChatGPT went mainstream, recruiting for key roles was largely happening via referrals or scraping LinkedIn for people at target companies.

When I posted jobs (for a moderately well known company name) on public job boards, we'd get 1000s of applications within the first week long before people were using ChatGPT to fill out applications. The majority of the submissions were absolute junk: People spamming their resume to every job out there, resumes from people completely unqualified (like tech support people applying for Staff Software Engineer roles), or mad-libs style resumes where people threw word salad into a PDF while forgetting to tell me what they actually did ("spearheaded initiative to reach across the company and leverage synergies, increasing company revenue by 23%")

It has been like this for a long time. ChatGPT only seems to have emboldened more people to switch to spamming resumes and sending vacuous applications everywhere.

I can understand having to respond if it's been 8 months, but this is a pretty big red flag that the company doesn't respect humans.
It’s a giant telecom company that already had some pretty big red flags around how they treat employees from my own network.

The best part is that she already interviewed there a while back. She got a backend interview and didn’t pass (she’s primarily frontend); not a big deal but she gets a message from a new recruiter about every 2-3 months about the same role. This time it was just AI doing it.

Someone is getting poor quarterly review results, and is looking to see how to replace them once the quarter is up; rinse and repeat
AI startup idea, CaaS, candidate as a service.
Yep, I just had this happen too. I'm not looking nearly that seriously but the position as described was nearly perfect. The AI phone call plus a couple other minor details (slight changes in the description/requirements) which could probably be easily cleared up with a human made me distrust the entire process and walk away.

It's incredibly annoying

Anyone have a text-to-speech AI to talk to recruiter text-to-speech AIs?
I love that idea. Record the messages and use that to fine tune your own language model to improve the chances of talking to a real person? If she does that and can’t get a job, the market is officially broken.
This is Real Genius but on the job market
Yeah but if it's a local minimum and legal too, nothing will change.
yes interested