Based on just observing him through the process, it seems like there's a combination of things that might keep younger people away from being an ATC. One might be the unfortunate politicization of government jobs at large, and especially how people like ATC's get affected during those pesky "government shutdowns". There's also, I think, the fact that being an ATC is more or less an "invisible job" to many people and especially college educated people. Also, the job itself can be stressful (but nothing like the horror story myths I've read about, apparently) and the exams to get in whittle down the herd pretty dramatically it sounds like.
That said, there isn't really a good path to ATC for those 18-22 either. The off the street bids require something along the lines of "4 years progressively responsible work experience" to even be eligible to start the process. The Collegiate Training Initiative (CTI) schools exist, but have not provided much meaningful benefit to graduates since the FAA's changes to the hiring process in December 2013.
Back then journeymen controllers in Chicago were making $150K.
I don't resent it though - it sounds like at the time CTI students weren't much more successful than off the street hires at the academy, nor were the different CTI schools very consistent in their quality and curriculum. I went to a good one, and was sad to leave as it was a ton of fun, but I've heard stories of some CTI programs being little more than a few lecture-based classes and a certificate.
The biggest two problems that immediately come to mind:
1) Current ATC communication with planes happens by human voice over radio. Wouldn't want to trust an AI's speech recognition over that medium, so you'd probably have to mandate some kind of pilot-ATC interface that all planes would need to carry. You'd get huge pushback from the community of small planes. Probably from commercial pilots as well, since they'd have to train on the new system, and it would likely add to their cognitive load, at least short term.
2) We've put a ton of effort into producing a good radar picture to let controllers know where planes are, but it's far from perfect and requires some intuition. Radar waves bend due to atmospheric conditions, and it's difficult/impossible to distinguish real targets from various forms of clutter. The heuristics we currently use are pretty amazing, but you still get things popping up on ATC's radar scope that aren't really there.
The tools ATCs use have evolved tremendously. Just three decades ago, not all major airports had ground radar for instance. Then there's ADS-B that's fully automated and does part of the job. The problem is that air traffic has increased a lot.
Even if automation goes much further, ATCs are needed for supervision of operation, compliance and backup in case systems fail. The evolution mirrors that of the job of pilots. Flight engineers have been rendered obsolete long ago, maybe copilots will be eventually replaced by a remote operators for backup at some point. Each ATC will be able to handle more and more traffic, as they already have, but someone's gotta be keeping an eye on the system.
Yes, so is actually flying a plane. However, MCAS let us see how badly things can go wrong following rules that were not well defined/understood/studied. Contributing to that was the system's method of knowing its environment came from sensors. If/when those sensors malfunction then the system is useless. At the end of the day, even if 100% of all instruments and sensors were to stop working, a human can still use its back up sensors (eyes,ears,brain) to continue flying where the auto system pretty much is game over.
At some point, they'll either need to find a way to accept more trainees, retain employees longer, or reduce services/throughput.
EDIT: I may be out of date. Perhaps this second wave already came and went with little impact. My math says anyone hired in 1981 would be past retirement age by now. Not sure how they handled it.