Question

First we had to learn to cough up the key words some Applicant Tracking System (ATS) wants to see when we submit a job application. Now we have to use ChatGPT to write our resumes and applications because employers are using ChatGPT to write job descriptions and postings. Where does it end? Why don’t they just hire the ChatGPT?

Nick’s Reply

recruiting-chatgpt(In this column, I will use ChatGPT as a catch-all term for all generations and competing products of the underlying technology, because this is not an analysis of the technology itself.)

An L.A. Times story tells how to write a cover letter and resume using the artificial intelligence of ChatGPT. The technology is credited by users with the skills of a professional coach or editor:

“The aspects of using AI to assist — it’s a tool,” job seeker Jesse said. “Imagine you had an expert next to you telling you how to get better… It wrote [a cover letter and resume] better than I ever could.”

Seriously? An expert?

Is ChatGPT an expert?

Human experts like Noam Chomsky describe these A.I. systems in The New York Times like this (emphasis added):

“Roughly speaking, they take huge amounts of data, search for patterns in it and become increasingly proficient at generating statistically probable outputs — such as seemingly humanlike language and thought….ChatGPT and its ilk [are] a lumbering statistical engine for pattern matching, gorging on hundreds of terabytes of data and extrapolating the most likely conversational response or most probable answer to a scientific question.”

“…machine learning systems can learn both that the earth is flat and that the earth is round. They trade merely in probabilities that change over time.

What game are we playing?

Employers and the Employment Industry at large are creating an increasingly complex game of recruitment advertising and hiring. Of course, now job seekers are learning to play the game by deploying ChatGPT against ChatGPT. (We warned you 7 years ago: Send a robo-dog to interviews.)

But how does this approach of gaming a system that is itself a game play out? Who gets hurt? Who benefits? Will Jesse get the right job? Will you?

What’s the outcome?

The proliferation of job boards and ATSes doesn’t seem to have fixed the talent shortage for employers, and it hasn’t fixed unemployment for workers. But we still use them, perhaps because we keep avoiding an outcomes analysis. Maybe it’s because automating it makes job hunting less painful, even if that doesn’t really work.

So let’s automate some more! But let’s check the outcomes, eh?

Does reliance on ChatGPT improve:

  • the quality of applicants or candidates,
  • the quality of hires,
  • the quality of a job match for the job seeker?

Or are we just getting better and quicker at pushing a square peg into a round hole before anyone realizes the damage that may be caused?

ChatGPT: Everybody can do it

Then there’s the problem of needing to use a cheat-checker to avoid getting caught cheating. Though Jesse lauded ChatGPT for cranking out good cover letters, saving him work, “there was one additional step involved: running the letter through online A.I. scanners that have popped up to detect A.I.-generated writing to make sure it passed the test in case companies checked.”

The L.A. Times story cites other job seekers who report that “tapping ChatGPT to write their cover letters was a no-brainer…. ‘A bot reads them,’ they said, referring to cover letter and resume-scanning software that many employers use to filter out candidates. ‘I’ll get a bot to write them.’”

Everybody’s doing it.

Interview the glove!

Kind of sounds like a nuclear proliferation treaty will be needed before real information about workers and jobs disappears in a mind-numbing hall of machine-learning mirrors.

Players on all sides of the Employment System have adopted a virtual process that creates avatars or surrogates to conduct the business of matching workers and jobs. Maybe another analogy is more apt: the old complaint about “washing your hands with rubber gloves on.”

Just hire the ChatGPT

Of course, what’s wrong with hiring someone who used ChatGPT to produce their cover letter, if this new employee will use ChatGPT to do their job, too?

Play this out, though: Who needs this new employee? The hiring technology could also be used to do the job.

Well, at least a lot of people other than Noam Chomsky seem to think so.

(Does anyone see what I see? ChatGPT is just the next level of keyword matching that drives sincere job seekers mad as they lard their resumes and job applications with strings of letters they know the algorithm is searching for. While the new tool is certainly more powerful — it will lard your resume for you — is it actually any different?)

What does the use of ChatGPT tell us about how the employment system works? If employers are going to hire based on auto-written cover letters and resumes, what does that tell us about how they assess job applicants? And of course, what does it tell us about job seekers who use ChatGPT?

: :

15 Comments
  1. Hi Nick
    I’ve been following you since 16 years ago. Your help was priceless.

    To answer your question: employers already hired ChatGPT. Countless jobs were already automated, and fewer ones were created in their place to drive the questions to chatGPT.

    Would anybody hire a job seeker that doesnt know how to “Google it” (the verb)?

    In a few years it will be obvious to replace the verb above with a new verb: “chatGPT it”

    Warm human regards

  2. In many different fields experts have played with chatgtp. In a disturbing number of those fields the experts are finding chatgtp not just collating data that sounds right, it ends up creating data that sounds right from nothing (I.e. lying and cheating).

    Many popular chess YouTubers (including US grand master Hikaru) have played chess against chatgtp. In these games chatgtp invariably makes up moves and rules and even respawns its pieces if they get taken. I totally makes stuff up.

    It does this for so many things. My son has a piece of guitar equipment and someone on the forum for the vendor posted how he asked chatgtp how to determine which particular version of the equipment someone has. Chatgtp came up with three reasonable sounding ways to determine this. All three were fabrications and not the actual way to do this.

    And then there’s the recent story where chatgtp makes up a story about Jonathan Turley committing sexual harassment that is totally untrue but has links to a nonexistent story about the fake charge.

    There is something truly disturbing in chatgtp willingness to fabricate.

    But when you think about it, and what Noam said about it, fabricating true sounding garbage is also a good description of what it does… which make is nearly perfect for fabricating resumes and cover letters!!! ;-)

    • @J: There are loads of worthy quotes in Noam Chomsky’s column that point out the fundamental, fatal flaw in ChatGPT. While it may make it easier to identify cancer in a liver (and I’m all for using it that way), it’s incapable of finding a cancer in, say, a government, or a liar in a debate. Because few people understand what a grammar is, they fail to see the critically weak link in ChatGPT.

      “Indeed, such programs are stuck in a prehuman or nonhuman phase of cognitive evolution. Their deepest flaw is the absence of the most critical capacity of any intelligence: to say not only what is the case, what was the case and what will be the case — that’s description and prediction — but also what is not the case and what could and could not be the case. Those are the ingredients of explanation, the mark of true intelligence.

      “True intelligence is also capable of moral thinking. This means constraining the otherwise limitless creativity of our minds with a set of ethical principles that determines what ought and ought not to be (and of course subjecting those principles themselves to creative criticism).”

      “Thinking” means more than storing billions of ‘facts’ and quickly finding patterns in order to predict what “probably” comes next.

    • As a loyal long time reader. I love this article. I laughed out loud.Once again, you hit the bulls eye.

    • I believe that HR also fabricates absolute lies, or garbage, about jobs it posts.

      If anyone is serious about improving employers’ ability to fill their jobs with “the right person,” try posting, as Nick suggests in another thread, what the employer wants accompliushed in the first 30, 60, 90, etc days.

      Job postings I continuously see on the several digital leed emails go on for pages, describing what no one could possibly unless they were 10 or 12 people. Or maybe ChatGPT.

  3. Welcome to the new world order.

    The book 1984 is something we all should be rereading right about now. Same with animal farm.

    You will own nothing, and be happy
    You will want nothing, and be happy
    You will question nothing, and be happy

    The current generation is already dumber than their parents, lazier than a sack of potatoes, and firm believer is anything that come from a “social media” platform. Why not trust this perverted “AI”?

    Anything to make the job easier.

    The older days were better, a sign said help wanted, you walked in and asked about the sign, if the manager/owner was not in the other employees were informed of what to do.

    Drones is what we will be.

  4. I foresee that pretty quickly applicants will use ChatGPT in a HireView session – turn on your microphone and voice recognition software, Hireview asks a question, the software feeds it into ChatGPT which provides a response, which you then read out loud to HireView.

    Or even better, get a realistic test-to-speech generator, and go get a cup of coffee while the computers talk to each other!

  5. I was thinking more on this. With chatgtp’s propensity to lie and make stuff up, the perfect job for it to replace and leave utterly destitute and unemployed is Politician!!

  6. Yet another poorly designed piece of software generated solely for the purpose of “cashing in”. Very tiresome but I guess I have better things to do with my time. Tennis anyone?

    • @Marilyn: I think there are important applications for ChatGPT’s amazing pattern-matching capabilities, but only for problems where the bottleneck is being able to process billions of patterns very quickly. If you think about it, a resume offers way too little data for ChatGPT to work on. ChatGPT may be able to sort a thousand possible candidates out of 2 million resumes, but it cannot “decide” whether a specific candidate is right for a job. The likelihood of being wrong with many of the 1,000 doesn’t really matter. Matching correctly just one candidate to one job is all that matters.

      • I was once a headhunter when my employer went out of business. I learned that the best decision for an employer is to hire the man who can do the job. (Maybe, the first man who can do the job.) Too much looking for “the perfect match” is an entire waste of time and resources.

        Also, if an employer wants to retain qualified employees, pay properly, don’t lie, be honest, etc. A SERIOUS question: Why do all (yes, ALL) employers lie?

  7. It’s bad enough when you submit an application and get a “no thanks” less than 24 hours after the application was submitted. You know that no human looked at the application or resume.

  8. (We warned you 7 years ago: Send a robo-dog to interviews.)
    I remember that article. Wow, was it really 7 years ago? Boy, time flies!!

  9. At what point are there no more people left who know or care about the objective truth/facts?

    Or so few people who know, that the other people lock us away or kill us off because we’re too much of a nuisance, what with insisting on reality and accuracy and logic and science and the like?

    It’s already pretty bad in the world for scientists and others who deal in evidence-based reality and for anyone who strives for accuracy and excellence.

    I absolutely am all for “laziness”, as I firmly believe in working smarter and no harder than absolutely necessary to get a good result. But I’m still a stickler for the GOOD part of the result! When does that become more of a liability than it already is?

  10. Sounds like ChatGPT has a bright future at one of the Big 5 consulting firms (assuming it can generate pretty PowerPoints).

    But seriously. Unless you are desperate (no shade—I have been desperate), don’t go work for an employer who forces you through ATS (including ChatGPT-enabled ATS processes). If the hiring manager is not taking hiring seriously, things are unlikely to go well for you.