Agentic AI. Will the Absurdity Never Stop?

I don't know if the folks at the consulting powerhouse PWC coined the terminology "Agentic AI" or if they merely picked it up elsewhere as a sub-topic of the many general AI conversations being had these days, but a PWC article is where I first became acquainted with this new terminology, thus I will aim my guns specifically at PWC, with the entire evolving AI industry also in my sights.

The PWC article begs the questions: Do humans themselves possess autonomous agency? And would not the article have been better served if the idea of autonomy had been kept out of the conversation? I ask these questions while realizing that "autonomy" long has occupied the discussions of workplace theorists and historians as that word (autonomy) has been applied to workers of the human variety, making it inevitable that the issue of autonomy would worm its way into AI discussions.

Regarding the first question, anyone trying to answer will, if thinking about the question more than superficially, eventually need to answer some other, deeper questions about who and what Man is—about whether Man was created or evolved, or some combination of both. And then a line of reasoning must be developed from one or the other theories (or both) regarding what it means for Man or Machine or Program to have "autonomy."  These are questions and issues that involve a deep dive into Ontology, Theology, and to a lesser extent, Philosophy. 

And while this methodology might be more work and require much more intellectual effort than many are willing to invest in exploring what seems like a side issue, we cannot let ourselves off the hook so easily once we have invoked the idea of either human or machine or program autonomy. Before we ask "how," we must first establish whether it—"it" in this case being the autonomy of "artificially intelligent" data-mining, data-managing machines (perhaps "mechanisms" might be a better word)—is even possible, much less worth our time and effort in considering the matter.

Is, or should be, Agency ever really autonomous?

Ontologically and Theologically speaking, humans have agency. But do they, or should they, have completely autonomous agency? And is not that autonomy complete or partial depending on who or what bestowed the agency upon them? Again, those are questions that arguably have not been settled, that is, we continue to debate the whole "human autonomy" thing. What makes us think, then, that the matter of autonomous "Agentic AI" can ever be settled satisfactorily?

A Historical Sidebar

All the technology ever invented / discovered was done so with any eye toward replacing human PHYSICAL tasks with technology-driven machinery. But now we have strayed into the realm of the metaphysical. We want to replace those non-physical things about us that make us uniquely human, but toward what end? If it is to reduce or replace the physical activities that arise as one of the results of exercising our intelligence and our ability to discover and invent, that is one thing. But it is something entirely different to attempt to replace (or claim that we have replaced) the uniquely human traits of intelligence and conscious action with "artificial intelligence" and "agency."

The Absurdity of "Agentic AI"

Again, what do PWC and others mean when they say that Agentic AI is "designed to act autonomously?" I would first compare AI to other, older technologies that have been introduced to the workplace, including all kinds of mechanical machinery, and I would compare with such inventions / discoveries as the PLC (Programmable Logistical Controller) board that tells the machine when and how to do what it does.

Production machinery and PLC boards are not, in and of themselves, "autonomous." Their very existence is dependent, and will remain so forever, because it is absurd and inconceivable to expect a machine to continue forever to "make decisions" (or, at least, the RIGHT decisions), without further, and often frequent, human intervention. Production machinery does not forever run itself. Neither does, or can, AI machines, programs and agents..