New labels warning

I’ve confessed before that I’m not a process control engineer, so I’m logically not one of those guys who can put a hand on a pipe or tank, or just listen and tell what’s wrong or not from how it sounds or feels. What I’ve learned over the years is to pay attention, mostly as I try to understand what Control’s many expert authors, contributors and other sources are describing, so I can help relay it to our readers.

I know I’ve missed more than a few important points, but I think I’ve also gotten good at knowing when useful details are being expressed or not. In fact, after editing so much text, and correcting so many misspellings—or fashionable overuse of dashes—I can see when writers appear to lose their initial focus and even interest in their subjects. It usually occurs in the last quarter or third of a long story. Errors and unnecessary words increase, vague statements grow plentiful, passive-voice language based on “to be” phrases multiplies, and first references already defined earlier get restated. These snags can trip up readers, and kill what’s left of their already short attention spans.

One consolation is that striving against bad writing helps identify a newer problem. Nebulous talk has been around forever, but at least there were many different and sometimes quirky ways to say little or nothing substantive. Now, much of the text I encounter seems to have an added glaze of beige foam and pink slime sameness, as if one breezy, AI-generated voice is repeatedly speaking for those who used to be talking.

Similarly, one of journalism’s traditional and somewhat sneaky directives for writing news and feature stories is, “Say it once, say it again, say it again, etc.” This low-expectation repetitiveness is right up AI’s alley.

When I wrote my first editorial for our quarterly Industrial Networking magazine (now Smart Industry) close to 20 years ago, I reminded its readers and myself that, “Talk is cheap,” especially if it doesn’t provide any useful information. I think I added that bloviating opinions were even less expensive, and that the Internet’s unlimited space made them cheaper still. Well, as we all know, the price has continued to plummet since then.

Luckily, even the shortest attention spans get bored eventually. So, what’s a hardworking AI agent, I mean, enterprising journalist supposed to do? How can all those readers, listeners and viewers be kept captive-ated? Change the names and wrapping, in this case, to protect the guilty. And pretend all the topics, devices, activities, dogs, ponies, smoke and mirrors being covered again and again are somehow completely new and different each time.

Get your subscription to Control's tri-weekly newsletter.

Yes, you’ve seen this show before. Low-power, point-to-point networks and fieldbuses were joined by Ethernet TCP/IP communications, which became the Internet of Things (IoT), the Industrial Internet of Things (IIoT) and a  bunch more variants. More data became big data, data lakes and other names I glazed over before retaining. Data processing became edge computing, cloud computing and, sweet Jesus, fog computing with a “cloud near the field” or something.

Most recently, someone likely thought artificial intelligence (AI) sounded too old-fashioned. I mean, even The Matrix movie, which highlighted AI, came out in March 1999, and AI was coined decades before that. So they put the prefix lipstick of generative AI (gen AI) and agentic AI on it, and sent it out on the street again.

Sure, some of these aliases represent software and other tools with distinct capabilities meriting different names, but most don’t. This whole ball of wax is simply software running on servers or modules linked by low-power, wired or wireless networks. Much simpler than it’s usually made to appear, right?

Unfortunately, the baloney spigot will never shut off because it’s so much easier to spew sparkling promises and drivel than it is to work on something genuinely useful that doesn’t need to trumpet as loudly. So, like any workplace hazard that can’t be mitigated, it must be caged, avoided and warned against.

I’d recommend what many sources in Control’s stories advise: use specific problems and detailed performance goals to guide your search for the few appropriate capabilities, software and other tools that prove they can solve and accomplish them. Flush everything else.