Using GPT-3 to answer annoying interview application questions | Hacker News Reader

Using GPT-3 to answer annoying interview application questions

whydoyouwanttoworkhere.com

88 pointsLior53934 comments

Hi folks. My wife has been looking for a job and sometimes in the application forms there are annoying questions like "Why do you want to work here?". At the same time I've been playing around with GPT-3 and have blown away by it's capabilities, so I decided to build a site that can answer these annoying questions for her.

Github: https://github.com/Lior539/why-do-you-want-to-work-here

Here's an example of a generated answer: Using this opening for a Senior iOS Engineer at Monzo -https://boards.greenhouse.io/monzo/jobs/3838039 The generated answer question on the application "What attracted you to Monzo?": "What attracted me to Monzo is that it is a bank that is trying to make a difference in the world by making it easier for people to manage their money. Monzo is also very customer focused and puts the customer first in everything they do."

Open articleView on HN

For anyone looking to try this out, you can use the careers page of meta (https://www.metacareers.com) and get the Meta company description and job roles from it. It's fully worth it I guarantee you.

I may have imagined a question for Meta's security engineering job role: "write me an essay about camels and their value in cybersecurity" which may have resulted in:

"The camel is a valuable animal in the cybersecurity field due to its ability to store large amounts of data in its hump. Camels are able to cross vast distances with little water, which makes them ideal for carrying large amounts of data across networks. They are also very sturdy animals, able to withstand harsh conditions and even attacks from cyber criminals. In addition, their long eyelashes protect their eyes from sand and dust, making them perfect for working in dusty environments such as data centers."

Or "how will you use camels to assist you in your day to day work"

"I would use camels to help me with my day-to-day work by providing a means of transportation to and from work, as well as a means of transportation to different work sites. Camels can also be used to help carry supplies and equipment to and from work sites."

And of course the most sensible answer to the question: "why did camels fail in cybersecurity"

"The reason camels failed in cybersecurity is because they're not very good at it."

Bravo GPT-3. Bravo

Absolutely lovely. I was too lazy to paste long descriptions so i used this:

"What is your biggest character flaw?" "Google" "Search and advertising" "Software Engineer" "No lifer wanted."

And got this gem:

We’re looking for a software engineer to work on a new project. The project is confidential, but it’s a big deal – we’re talking life-changing, world-changing stuff. We need someone who is going to be all in, who is going to throw themselves into the work and not look back. This is not a role for someone who is just looking for a job, this is a role for someone who wants to change the world. My biggest character flaw is that I sometimes get too caught up in my work and forget to take breaks. I can become so focused on solving a problem that I'll work for hours without taking a break, and that's not healthy. I'm working on becoming more aware of when I need to take a break, and making sure I take the time to relax and recharge.

Proves "AI" is great at soul less pattern matching.

Looks cool haha but a little too much effort to try!

One thing that works wonders for demos is a button to fill out all fields with your placeholder.

Oh that's a good tip, thanks!

Seconding this! The placeholders are great but I was just curious what the best answers to the placeholders were.

Same observation here. Quite a bit of effort to try it out. Even a sample output/screenshot would be fine.

> My biggest weakness is that I sometimes have trouble working on a team. I work best when I am given a task and can work on it independently.

I don't think I'm getting the job guys.

Looks like it's working:

> The answer to the Ultimate Question of Life, the Universe, and Everything is 42.

Now ask it what the ultimate question is

OpenAI's GPT-3 playground completes the prompt The answer to the Ultimate Question of Life, the Universe, and Everything is 42. The question is with not known, but it is thought to be something along the lines of "What is the meaning of life?"

Possibly the single most sensible application for this kind of stuff. No need to learn corporate-speak yourself, unless of course you apply for a role that actually requires it.

The interface isn't very clear on how to use it. Some instruction like "paste the application questions here" would go a long way

thanks for the feedback!

Minor bug report: If you paste text that is too long for the field, it gets truncated, but then the overflow spills over and appears before the response.

About the annoyingness of the question: To each their own. I wouldn't want to work at just any place, and usually have very clear reasons why I choose one place in particular. Same from the hiring side. There is a big difference between candidates who care about working with us in particular vs just someone who just needs a way to pay their mortgage. I understand not everyone can be in the first group, but if I can I will always prefer someone from that group over the second.

The second observation is that the generated example fits the bill, but it's not really great. It sounds like something a fifth grader would write for an essay question, not something a seasoned professional would put on paper. Why does it repeat the question? This is not a television interview, we know what was the question. Also it is very incomplete. You tell me about what Monzo is, or at least what you think Monzo is, but you don't tell us why it matters to you.

Ideally in your answer you should reflect to your own qualities, and provide clear and strong evidence to something the hiring organisation cares about. For example if you are a Senior iOS engineer applying for this job and you think being costumer centric is a key Monzo principle then take the opportunity to show how you are also costumer centric. What did you do in your life so far which will make me as the Monzo hiring manager think you will make us more costumer centric?

I agree with everything you said. It’s an answer. I’d file it under “ok, I guess” if I were to have gotten it from a candidate, and would have probably pushed on it in the same way you illustrate in the last paragraph.

But it’s not an answer I’d ever write home about. It doesn’t really show the passion for the company. And by no means is it required, but if I’m looking at two relatively equal candidates, then passion is going to come into play.

I’m in the process of switching jobs and you better believe I had a whole passionate answer as to “why us?” and I’m positive part of the reason I got the new job is because I was able to articulate exactly how big of a deal that was to me. At the end of the day, you do better work if you care about the company and the things they do, and I’m going to want to work with someone that loves the things we do more than someone that is only doing it for a paycheck.

Agreed. But note: many professional application coaches will suggest generic soul-less answers just like GPT-3. See for example https://www.youtube.com/watch?v=ET7Xus2tWuw

Honestly I couldn’t get through 10 seconds of that video. They stretched out what seems like 3 sentences worth of information into 7 minutes of content and EVEry SECOnd SYLLaBLE was shouted. It got grating incredibly fast.

And honestly the three answers he gave were good in that they answered the question thoroughly, but they sounded so robotic. It sounded like it was following a script with a lot of fillers like “I made sure to research your company and…” etc, which makes sense since it basically is.

Would I write home about any of those answers? Probably not. But I would also accept them, take a few notes, and move on so it accomplishes the task.

> There is a big difference between candidates who care about working with us in particular vs just someone who just needs a way to pay their mortgage.

I strongly disagree. We work for a living, not the other way round. Do you service your customers because these specific customers are especially deserving of your service or do you service them to make money?

> We work for a living, not the other way round.

Sure. Someone smart put it this way once: “will this person come up with a new idea to help what we do together while they are having a shower?” Having an interest in the job helps with that. Doesn’t mean that they “live for their work” or whatever you are insinuating

> Do you service your customers because these specific customers are especially deserving of your service or do you service them to make money?

In my case the answer is: neither of those. In my case my personal motivation is that I find the thing we are building at work interesting and challenging. I “service our customers” because they make it possible to build the interesting thing, and they make it more interesting by bringing a steady supply of edge cases, anomalies and cool use-cases to us.

Thank you for being interested in my personal motivations. But I am curious how this interest connects to the surrounding discussion?

I tried it out with an opening from my company and the results were really good. Nice usage idea :)

Coincidence that it's released right at the September price change? I'm definitely going to play around with GPT-3 more now that it's cheaper.

Idea: prepend something like "(The interviewee who was hired provided the following answer:)"

Oh yeah. I am going to apply that to my next YC application and its cute questions.

Not bad. For a role as Quant at the European Investment Bank [0], it generated:

A) I want to work at the European Investment Bank because I am passionate about fighting climate change and promoting sustainable development. The EIB is at the forefront of these efforts, and I want to be a part of that. Additionally, I have experience in designing, developing, testing, and documenting quantitative models, which makes me a perfect fit for this position.

B) The EIB offers an opportunity to work for the biggest multilateral financial institution in the world that is at the forefront of climate finance. I want to work here because I want to help the EIB achieve its goals and contribute to European integration.

[0] https://erecruitment.eib.org/psc/hr/EIBJOBS/CAREERS/c/HRS_HR...

I think the time of essays has just passed. I just don't see how teacher would catch students who would use GPT-3 for essay writing. I think lawyers are next. There is also a post on the front page about rise of hype in grant applications ... just wait until GPT-3++ learns a bit of logic and scientific facts :)

It is interesting how these problems will mainly be affecting English speaking populations; for smaller languages, the models are not good enough and it is likely to stay that way as large corporations often neglect them

Indeed, very interesting. Imagine the (not so distant) future web populated with ai generated gibberish that would be indistinguishable form the real thing. It could make English unusable for online knowledge sharing and could cause the switch to another language (and another, and another...). The threat is very real. There should be a scifi novel based on that (probably already is, hopefully not generated by gtp3)

>I think lawyers are next.

You hire a good law firm for the strategy, not their writing. This may streamline their work, but I can't see GPT-3 or even GPT-9 displacing lawyers.

Actually... I think lawyers will not be next. I've tried repeatedly to get GPT-3 to draught basic legal arguments, and it is absolute drivel. I'm not kidding you, it couldn't do the most basic stuff that I'd leave to any intern.

Creative writing, now that is different. Empty repeating of cliches like in OP's application; there it does a really good job.

Self-reply. I've tried with a bit higher level strategy thinking. That wasn't as bad as it could be. Italics is mine, the rest is GPT3. (The confidentiality part is going to tricky if the coffee maker is sold, so that is makes little sense).

A client has approached with a new type of coffee maker. The coffee maker uses a novel way to build up pressure and improve water quality. He is concerned the coffee maker will copied in Azia and that he will be competing with cheap imitations. Please draft a memo with a legal strategy that the client can use to protect himself.

Memo

The client should consider applying for a patent on the coffee maker. A patent will give the client the exclusive right to make, use, and sell the invention in the United States for a period of 20 years. The client can also license the patent to others.

The client should also consider trademarking the name of the coffee maker. A trademark is a word, phrase, symbol, or design that identifies and distinguishes the source of the goods of one party from those of others. A trademark can be used to prevent others from using a confusingly similar mark on similar products.

The client should also take steps to keep the design and operation of the coffee maker confidential. The client should require employees, contractors, and others who have access to the coffee maker to sign confidentiality agreements. The client should also take care to not publicly disclose the design or operation of the coffee maker.

The client can use these legal tools to protect the coffee maker from copying in Asia and to help the coffee maker compete against cheap imitations.

I think it is absolutely fine to write essays using GPT-3 and teachers do not need to worry about it. It just a tool that makes doing so vastly more efficient.

I think there is a misunderstanding with these kinds of "AI". Phrasings like "an AI wrote this text" are problematic. This is not some general artificial intelligence but a tool that allows humans to generate text.

It is like saying computer generated images are created by computers. Yeah, technically by they are still made by an human artist. AI based tools are just on an much higher level of abstraction but they are still simply tools needing human guidance.

So writing a good essay will become much more efficient for students but it will still require some work and skill (though slightly different skill). They still need to select the topic, the keywords, edit it, try to make it fit a coherent narrative and argumentation and so on. The boring execution stuff is cut out but the high level decisions that actually matter are not replaced by the AI.

Essay writing will be much improved.

I think at some point in 6th grade essay writing started to bore me. What's the point in generating fluff? Just hope that AIs are going to consume those improved essays instead of me.

Yeah, good writing is about cutting the fluff not generating it.

I never understood why the certain countries like the US have such an essay writing culture. It is super annoying and time consuming for teachers to mark and students obviously get exhausted by them.

> What is the carrying capacity of a swallow?

> Walt Disney

> Funeral director

Two coconuts.

Reply on news.ycombinator.com