Another Big Sin: Overreliance on Automation

Even before ChatGPT set things ablaze, digital media relied too heavily on robots minding the store

“This doesn’t even make sense.”

I’m processing what’s been done to my resume and trying to pick my jaw up off the floor while giving some sort of constructive feedback.  I’m failing at all three.

The situation was this: Looking for a full-time job for the first time in almost two decades was a slog.  I was following the old process of, you know, sending someone your resume when they posted a job I thought I’d be good at. This approach was yielding precisely nothing in the way of interviews, calls or anything that could be construed as interest on the part of the employers I was trying to engage.

I knew what was going on, having recently tried to launch a recruiting business with three industry friends.  I had put myself in charge of quizzing my partners on their pain points and desired features when it came to Application Tracking Systems (ATSes) and how they managed the krillions of applications every public job listing elicited.  I looked at a wide range of them and had selected a partner.  Unfortunately, the business never got off the ground for reasons I won’t get into here.

Lessons learned from a failed startup

The big lesson learned from the failure to launch was that employers were dealing with floods of applications, a natural result of the frictionless “one-click application” processes that many job boards and B2B social networks had offered to job seekers.  Of course, the ATSes were claiming to use Artificial Intelligence to filter through resumes and pluck out the most qualified applicants.

It was suffering from the same thing digital media has suffered from since the first sleepless night your adops person spent uploading campaigns to the adserver – an overreliance on automation.  This is another of digital media’s Seven Deadly Sins.

It’s certainly not unique to digital media.  Thanks to the popularity of ChatGPT, Midjourney and other “AI” plays, automation is disrupting nearly everything today.  And its proliferation isn’t necessarily a problem in and of itself.  The problem is what we’re willing to give up when we automate digital media tasks.

But back to my resume…

“Word Salad” would be the best way to describe how my resume looked after several rounds of revisions.  Along the way, my criticism of its metamorphosis was that the task my resume needed to perform was bigger than merely making it past the ATS.  A human would have to read it, eventually, and I couldn’t see how they wouldn’t be confused. 

The entire exercise was beginning to remind me of pre-Google search engine optimization, where keyword-stuffing played an outsized role in impressing the underlying tech of AltaVista, Lycos and HotBot.

How did I end up here?

Ironic, isn’t it?  (I mean, Alanis Morissette ironic, not Quint succumbing to the shark after surviving the Indianapolis disaster ironic.)  Here we have someone who understands not just digital media’s overreliance on automation, but how applying for a job within the digital sphere is affected by crappy ATSes, and he’s still struggling with how to put a resume together.

In the end, I went back to my old resume and put a ton of energy back into personal networking, which is getting much better results, thankyouverymuch.

The metaphor works. I swear.

The reason I’m even continuing with the metaphor, though, is because the ATS problem is actually two problems in one.  And they mirror the issues we have in digital media:

1. We look to AI to solve problems because the investment community rewards us for it in the form of higher valuations (and multiples when we sell). 

2. Even if our tech isn’t AI, but mere algorithms or machine learning.
   Automation solves for oodles of tedious work, but it often does so at the cost of insight.

I’m not suggesting we should return to the days of manual reporting or having humans review thousands of resumes for every open position.  What I am suggesting is that when we solve for the busy work, we don’t also remove our ability to deliver the insight that comes from familiarity with the tools involved and how they do what they do.

Automation can drive disconnection from the work

In the formative days of my former agency, we did some consulting work for an ad management company.  Chief among their issues was the notion that the more they did to automate tasks for adops and media people, the less those people were apt to understand how the system worked.  What was a classic SaaS problem of “80 percent of my users use 20 percent of the software’s features” was turning into a 90/10 split.

What was missing was the notion that automating all the work was causing adops and media people to divorce themselves from the source of their insights.  That overreliance on automation disconnected people from their understanding of how the underlying systems worked, and simply hastened the process of moving on to the next client campaign.

Human oversight is key

This is not unlike letting a crappy algorithm filter thousands of resumes directly into the trash because the words “growth marketing” didn’t appear enough times in the copy.  You might get to the end hire sooner, but you lose countless candidates who would have been better at the job, not to mention your understanding of how the right people end up on a short list for a hiring manager.

Now, of course, in the hiring sphere, we have a cottage industry of so-called experts who will jazz up your resume for a fee.  The goalposts have shifted from getting the job offer to getting past the AI.  Just as in the digital media space where the goalposts have shifted from achieving business objectives to achieving odd surrogates for those objectives.  It’s all part of how humans look at the rigid rules of technological systems – and exploit them. 

But that is a topic for next week.