Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A new obstacle to landing a job after college: getting approved by AI

cnn.com

Post reply on HN
Print view View original

91–100 of 130 posts

Re: A new obstacle to landing a job after college: getting approved by AI

#91

Post by john_moscow »

While I agree that using AI to rank the candidates based on their responses to pre-recorded videos is dystopian, I would like to point out that it only works in the environments where:

a) The supply of candidates greatly outweighs demand.

b) There is no clear way to assess the individual candidate's impact on the business' ability to make money, and the employees are essentially commoditized.

These environments have been known to be toxic for a while: filtering people based on bullshit 20-page personality questionnaires, managers hiring the most desperate ones who would tolerate the most humiliation from them, while every part of the process trying to a wink-wink nudge-nudge to others in order to land their nephew on a cushy position.

While AI definitely starts another round in the battle, I would advise anyone who's sick of this bullshit to focus on the economic output of their prospective position. If a company is looking for an AI engineer that can halve the size of their inference model without losing more than 1% of accuracy, they won't waste time scoring your story about your weak sides. They will look at your thesis and hire you if you already did something similar. The problem is that most of the jobs these days are about sitting quiet in a friendly team and smiling when the CEO shows off the team to the next round of potential investors. But those jobs won't give you much growth, and will end your career once you don't look young and cool anymore.

Re: A new obstacle to landing a job after college: getting approved by AI

#92

Post by bencollier49 »

retrac wrote: post #64
bencollier49 wrote: post #47

Most probably unusable in the UK, as the AI won't be able to explain why it rejected a candidate ("why did you score them low for team working?") and the candidate will take the employer to the cleaners. God only knows what hidden variables it'll start discovering and exploiting. My guess is it'll reinforce well-known prejudices.

I'm quite hard of hearing. I wonder what the algorithm's analysis of "facial expressions and the tonality of the job applicant's voice" would supposedly reveal about me. Probably something actionable under the nondiscrimination laws here, thankfully.

"I didn't like the look of him" certainly isn't an acceptable answer, and it's literally the answer in this case, always.

Re: A new obstacle to landing a job after college: getting approved by AI

#94

Post by perl4ever »

thaumasiotes wrote: post #61
perl4ever wrote: post #59

Earlier quoted context omitted.

According to Purdue, the first computer science department in the US was established there in 1962, so of course in 1960 IBM wasn't that selective - they couldn't hire CS graduates. Maybe math majors. Also, I don't know when CS degrees became common among programmers, but it didn't happen right after they were invented, I'm sure.

Did anything in the material I quoted, or in the remainder of either article, refer to CS degrees or CS graduates?

It's an example of conventional attributes which didn't exist then because the whole field was unformed. It seems like that would inhibit specialized aptitude tests - if you're hiring math or whatever majors back then (not that it doesn't happen today) you can't ask them about linked lists or sorting algorithms.

Not to mention, a vast number of people who could be good programmers wouldn't have had access to a computer before a first programming job, even if they had a college degree. Today it's much safer to assume that if someone is any good, they will have been exposed.

Re: A new obstacle to landing a job after college: getting approved by AI

#95

Post by thaumasiotes »

perl4ever wrote: post #94
thaumasiotes wrote: post #61

Earlier quoted context omitted.

Did anything in the material I quoted, or in the remainder of either article, refer to CS degrees or CS graduates?

It's an example of conventional attributes which didn't exist then because the whole field was unformed. It seems like that would inhibit specialized aptitude tests - if you're hiring math or whatever majors back then (not that it doesn't happen today) you can't ask them about linked lists or sorting algorithms. Not to mention, a vast number of people who could be good programmers wouldn't have had access to a comput…

But nobody's talking about interviews that ask for conventionalized CS material. The point is that the interview questions are more difficult. They aren't especially more computer-focused.

Re: A new obstacle to landing a job after college: getting approved by AI

#96

Post by throwthisaway2 »

danpalmer wrote: post #30

My brother has just finished a masters programme and is starting to apply for jobs, he's coming up against a lot of these. He's smart, motivated, very hard working, and has the necessary credentials on his CV, but when talking to him he's quite monotone and low energy. Because of this, he's failing these automated video interviews as he looks like he's not a "team player" or "personable". While interviewing practice…

Because he is not a team player... sounds they have done machine learning in who is good a "good ole boy". That is very soohiscated for the frat boys. Hats off.

Time to evolve. In the end they need someone like your brother, and your brother is better off working for someone else. But maybe your brother can game the system that is gaming him.

When software eats the world engineers are the muscle that clamps the jaw. Its not surprising that measures are being taken by those that will be eaten.

I think the midlife crisis is worst for engineers because we have to come to terms that certain people have cheat codes in our logical system. Its typically called nepotism.

Re: A new obstacle to landing a job after college: getting approved by AI

#97

Post by chrisseaton »

danpalmer wrote: post #71
52-6F-62 wrote: post #62

Earlier quoted context omitted.

I had never heard of these video interview analysis methods prior to this thread. I think it’s at the very least way too early to rely on this and at the worst just... terrible. AST’s are already notoriously bad and this is just a whole new level of just plain bad. Time is precious, sure, but so are human relationships if time means anything. It’s like nuance was completely disregarded in the pursuit of corporate con…

They do sound terrible and I hope to never encounter one, but my brother made a good point: the NHS get something like 12k applications per week, they have an average of 25k job openings each month. They apparently can't deal with that manually, which does make some sense. I personally think they _should_ handle it manually, and if the system breaks we should fix it in another way because this way sucks, but I'm not…

> the NHS get something like 12k applications per week, they have an average of 25k job openings each month

That’s just a couple of applications per job. How can the line managers possibly not have time to review just a couple of applications per person they’re apparently going to be leading?

Re: A new obstacle to landing a job after college: getting approved by AI

#98

Post by perl4ever »

thaumasiotes wrote: post #95
perl4ever wrote: post #94

Earlier quoted context omitted.

It's an example of conventional attributes which didn't exist then because the whole field was unformed. It seems like that would inhibit specialized aptitude tests - if you're hiring math or whatever majors back then (not that it doesn't happen today) you can't ask them about linked lists or sorting algorithms. Not to mention, a vast number of people who could be good programmers wouldn't have had access to a comput…

But nobody's talking about interviews that ask for conventionalized CS material. The point is that the interview questions are more difficult . They aren't especially more computer-focused.

I wasn't asked more difficult questions in 2007, when I got my first programming job. Several years on, we had some discussions about what kind of questions to ask candidates. One place I applied to more recently had a flowchart reading test that must have been from the punched card era.

So I don't concede the point, anyway.

But, the examples above definitely do compare more and less computer focused things.

Re: A new obstacle to landing a job after college: getting approved by AI

#99

Post by BickNowstrom »

drongoking wrote: post #23

Whenever I see AI used for something like this, right after thinking "our dystopia is here," I wonder how it could be gamed. An experienced interviewer can tell when they're being told what they want to hear; I suspect an AI could not. The AI ends up rewarding preparation and acting skills. And when you make too many hoops for people to jump through, eventually you find you're just selecting for those who are good at…

Students in South-Korea have a strong prep culture, and are already trying to game these types of interviews. It is much harder than gaming skill-based tests ("Don’t force a smile with your lips, smile with your eyes.").

> The AI interview is too new, so job applicants don’t know what to prepare for and any preparations seem meaningless since the AI will read our faces if we make something up.

I strongly feel the market will work itself out: If AI interviews bias companies to hire hoop-jumping and actors, in half a decade, other, more traditional hiring companies will eat their lunch. If the AI interviews only deliver top notch candidates... it is working as intended, and who am I to complain about my own lack of prep or acting skills? Not like whiteboarding or being able to quickly think on your feet without access to StackOverflow is biased against me already, and companies who use these for hiring seem to be doing just fine.

https://www.reuters.com/article/us-southkorea-artificial-int...

Re: A new obstacle to landing a job after college: getting approved by AI

#100

Post by perl4ever »

perl4ever wrote: post #60
jacobwilliamroy wrote: post #22

I think job applicants should have to compete with eachother in a craps tournament. Losers are knocked out of the pool until the number of applicants matches the number of open positions. I also think this is how we should choose the president.

There's an old joke that the first cut is or should be made by randomly throwing away a subset of applications, because you don't want to hire someone who is unlucky.

But seriously, I think it is probably a good idea, if you have to throw out 90% of applications to start, to do it randomly if the alternative is a dubious statistical or ML-derived model. It would make me feel better from the point of view of hiring or as a candidate, not to have the risk of an unforeseen undesirable bias.
Post reply on HN