LinkedIn is going all in on AI for recruiters and job seekers, and it's not good for anyone, except…

When the news hit last Tuesday, people started freaking out immediately and all my devices lit up. I had been driving, so when I finally stopped at a gas station, I checked my phone.

"LinkedIn launches its first AI agent to take on the role of job recruiters."

Aw man, that's it? I thought a celebrity had died.

In all seriousness, LinkedIn's play to be the AI black hole in the center of the tech job universe is the most obvious move in the tech world. LinkedIn has already been crowned as the hiring hand of God, while also awkwardly existing as the social networking arm of Microsoft, who themselves are a deeply-entrenched partner with OpenAI.

So this shoe had been dropping like a filthy Doc Marten on a freshly-mopped kitchen floor.

Yeah, that's gonna make a mess. Mom's gonna be pissed.

But let me be clear. This is not good. So not good. Unspeakably not good. I usually hedge on these kinds of proclamations, but in this case, I know it's not good. You know it's not good. Hell, LinkedIn probably knows it's not good.

Here's why.

OK, Let's State The Obvious

This helps no one except LinkedIn, and they're not doing a great job of jazz-handing that.

AI has done nothing but make the job search infinitely worse and it's done that quickly — a cascade of unintended consequences that have been falling onto the heads of recruiters and job-seekers alike for months, if not years. Much more so on the job-seekers, especially lately.

Furthermore, it's become obvious that — unintended consequences or not — the tech job market has rapidly become the most visible example of what I call the AI redundancy loop.

This is when AI is used not to assist humans, nor to execute tasks in situations where humans can't — both good use cases — but rather to replace humans in executing tasks that humans were already executing poorly. Then, when the results come back just as bad or worse than when those tasks were executed by humans, more AI is thrown at the problem, resulting in more tasks executed poorly, and so on.

Simple loop: Start using AI to screen candidates, which results in poorly matched candidates, which results in candidates having to apply to more jobs to get seen, which results in candidates using AI themselves to get past the AI screening, which results in an avalanche of completely mismatched candidates for every single posted job, none of which get filled.

Tell me I'm wrong.

What Your New LinkedIn Boss Overlords Will Look Like

And then finally, I guess, LinkedIn has been both the entry and exit point for the majority of the swirl in this redundancy loop for the last several years.

Let's tear apart the details of their new toy as documented by the fine folks at Techcrunch, the article referenced in most of the texts I got.

Short Story: The LinkedIn Hiring Assistant is being described as the one-stop-shop for all company hiring needs — job descriptions constructed from "ingesting scrappy notes and thoughts," sourcing the candidates, and even "engaging" with said candidates.

That last one. That's your nightmare, right? That's when you go Michael Bolton (no, the other one) on your laptop.

Anyway, the Hiring Assistant is being touted more for its AI than its HR, because this is LinkedIn's first AI agent — the next step up from LLMs — which incorporates expert knowledge to create a specific and dedicated kind of interaction. So not just customer support chat, but customer support chat that knows everything about how the product works.

The "product," in this case, being you and your career.

Why This Will Succeed

I would have put "Succeed" in quotes but I'm trying to cut down on being a jerk.

Also, remember, this is not for you. This is not for the recruiters. This is for LinkedIn.

The clear comp is Google and the advertising market. Remember when digital advertising was sold as a way to produce both more targeted ads for the consumer (job seeker) and lower cost of conversion for the advertisers (hiring company)?

Now, how long did it take you to realize that digital advertising is a dead game, because I figured it out in 2020, and I was like five years late. Ten?

The strategy is something like: Make a thing terrible from the outside until it's weak enough for you to own it. This also seems to be AI's general MO. Here, it's the difference between Workday's horrible user experience and LinkedIn's Easy Apply button. Results may vary.

So Hiring Assistant actually does have a potential champion — I'm speculating. From what I'm told, there is a cabal of recruiters and HR people, a "type," that are wholeheartedly embracing AI the way a software developer would embrace a new cutting edge coding language. I mean, it makes sense. The HR crowd (and I love them) just doesn't get a lot of new tech toys, and when they do, they're often hard to use and soul crushing.

Hiring Assistant, with its foray into AI agency, gives them a reason to tell unliked coworkers that they're screwed.

You know, like "My dude, you don't own any crypto? Good luck buying anything with your 'paper cash' in 2022."

Yeah, maybe I don't love all of them. I definitely love some of them. Some of them sure as hell don't love me. More on that coming.

Why This Will Fail

I don't know if it was the first time I laughed reading the Techcrunch article but I definitely laughed the hardest when they noted LinkedIn's previous forays into AI resulting in some "surprisingly accurate connection recommendations."

For a period of over a year, LinkedIn was recommending I connect with people who had the same name as people I was already connected to.

Let me make sure you understand that.

I have an existing LinkedIn connection with Greg Smith, a software genius I used to work with and now a CTO in New York. LinkedIn would recommend that I connect with Greg Smith, an insurance agent from Duluth, MN, because his name is also Greg Smith, and it heard I liked Greg Smiths.

I don't think they're still doing it, but one of my current recommendations is for Anthony Volpe, which got me super excited until I realized it wasn't the shortstop for the Yankees. I'll call that coincidence.

Also, from the article, "The product includes the ability to upload full job descriptions, or just note what you want it to have, along with job postings that you like the look of from other companies or roles."

My italics. One: This is another "Please Pay Us To Plagiarize For You" AI use case. Two: I already spent a whole column on why this is a terrible idea. TL;DR: It's just another notch in the AI redundancy loop.

But ultimately, what LinkedIn has done so far with AI — and this is in line with like 90 percent of current AI use cases and not coincidentally 100 percent the AI problem — has been firmly rooted in what I'm now calling the "Nobody Asked For This" AI use case, including, in this case, AI "learning coaches, marketing campaign assistants, and candidate sorters; writing and job hunting helpers; and profile refreshers."

Profile refreshers.

This is who we're trusting with our future careers.

What You Should Do About It

Nothing. I don't think this is going to work, but even if it does, well, read on.

Before tossing an enormous word salad, Hari Srinivasan, LinkedIn's VP of product, said in an interview, "It's designed to take on a recruiter's most repetitive task so they can spend more time on the most impactful part of their jobs."

Which part, exactly? If AI is writing the JD from "scrappy notes" and then "candidate sorting" (my quotes, for snark, can't help myself) and then engaging with the candidates, which part is the recruiter supposed to make an impact with?

Wow, maybe those unliked coworkers really are screwed.

But here's the thing. I was saying the same thing about repetitive tasks and impactful parts back in 2011 when I was developing NLG (now Gen AI) with Automated Insights. Back then, we weren't getting rid of writers, we were just getting rid of the bad, phony data scientists. And even back then, those phony data scientists, instead of learning to get better at their job, they just embraced our shit.

The companies that lean into this are not the kind of companies you want to work for. The candidates they hire will not be the caliber of candidates that will help them succeed. The redundancy loop never results in a fix, it usually just gets scrapped after making a huge mess on the kitchen floor.

It might be a long wait, but I say wait it out. In the meantime, hop on the "Networking and Making Real Human Connections" bandwagon before it gets too crowded.