American View: Why the jobs market is devolving to robots lying to other robots

I assumed, at first, that this was because Eduardo’s company was staffing up to service a huge new contract. That turned out to be wrong. The problem, my mate explained, was that his company’s applicant pipeline was stuffed to bursting with unqualified candidates.  

For context, Eduardo works at – worked for? is reworking for? works again at? – a high-tech manufacturing company. They need qualified and certified electrical engineers. Traditional machine operators won’t do. You might expect that job seekers reading this company’s job postings would self-select out when the posted role requires skills and qualifications they don’t possess. Eduardo insists that’s incorrect: over ninety percent of the candidates that have made it through his HR department’s screening process turn out to be completely unqualified. Eduardo has had to implement a high school level screening exercise at the start of every interview. He’s empathetic with the applicants’ plights but has no choice but to boot them out of the queue.  

Here’s why I think this is happening: first, several states in the U.S.A. demand that people getting unemployment cheques complete a minimum number of “job activities” each week. I think it’s three for Texas. These can be applications, interviews, or (in some states) training. That weekly requirement will motivate some unqualified people to shoot their shot at a prestigious and high-paying manufacturer … but I suspect that’s the least likely culprit.  

Second, pseudo-AI services are the hot, new, “must-have” competitive edge in the American jobs market. Theoretically, these services can speed up your searches and applications. Some tools aggregate tons of postings and use machine learning to optimise your CV for screening. Others add the ability to have an AI agent complete the online application for you. Still others will automate the entire process: you upload your data, then an agent will find comparable positions, draft a new custom CV specifically for the job specs, and apply for you. You wake up every day to see a new list of completed applications.  

I tried one of each service to see how well they worked, and I’m pretty sure this is why Eduardo is getting so many unqualified applicants.  

I found the ML aggregator to be utter garbage. Even with my full CV uploaded, over 90% of the open reqs the service sent me every single day were for security guard positions. Sure, there was some range; the reqs included mall cops, anti-shoplifting detectives, building guards, and actual police officer positions. Whoopie … that ain’t what I do. I’m not sure if the ML agent got tripped up by not understanding the “cybersecurity” entries in my work history or assumed that anyone with a military background was only suited to stand still and glare all day.  

On a side note, even after I set that first site’s filters to never show me any posted req that paid below $20/hour, it kept sending me minimum wage gigs. That’s $7.25/hour in Texas (about €6.36). If your “intelligent” agent can’t do basic math, I won’t be using it.  

Next up, the pseudo-AI service that tries to automate the application process worked slightly better, but honestly it wasn’t any faster or easier than just doing the entire process manually. There were so many errors to fix that it was more frustrating than helpful. Nice idea, but nowhere near ready for use. Worse, it often changed words from me CV to better match the language in the job posting but did it in such a way as to completely misrepresent my qualifications. 

Finally, the third site – the “let us find a gig and apply for you” site – was a horror show. I let it run wild for one day and immediately shut it down. In just six hours, this service found and applied for me to six gigs: 

• The first and sixth were for other shady pseudo-AI startups doing AI stuff that I’m not at all qualified to do 
• The second was a partial fit but got instantly rejected. Probably because something in the “automated” application was glaringly wrong 
• The third involved building vocational training for a construction company. Do-able, but a very weak fit  
• The fourth was posted for less than half of the absolute minimum salary I’d specified 
• The fifth was a nice in-office gig at a good rate … located a thousand miles away from where I live. I’d specified only to consider gigs in the DFW Metroplex or fully remote 

I’m pretty sure this is why Eduardo’s company is getting barraged with unqualified candidates that somehow make it through automated screening. Sure, these “AI assisted” or “AI accelerated” services seem helpful, but they’re not doing their customers any favours by wasting Eduardo’s (or any other interviewers’) time. Making everything worse, out-of-work folks like us can’t get in to the interview queues when so many “duds” have packed each queue as soon as an open req is posted. It’s like a well-meaning DDoS attack. 

I’m not sure how best to fix this situation. This goes beyond just the “teething problems” one expects from a new tech-based service. This is becoming an active menace. Still, I empathise with job seekers who will try anything to find work in our dying economy. Desperation can drive an otherwise honourable person to abandon their personal and professional standards if it means being able to afford food and shelter. I suspect this mess is going to get much worse before we get a handle on it.