Fallacious Ideas About AI Writing

The great flood of generative AI use has led to some writings conventions or styles getting assumed to be indicators that something was written by AI rather than a person.  Among these things is usage of the em dash—like this one!  I know I am writing this myself instead of having AI generate something with my textual inputs (and I will never utilize the latter for my blog).  I have used em dashes in my blog posts for years and years, in fact.  The evidence is there in the form of many previously published posts.

However, the foundational reason why something is or is not true is because logical requires it.  Concrete examples are secondary at best, examples of something that is already true or possible because of pure logic.  Whether an em dash has been placed by a human or by an AI is hardly as important an issue as all sorts of basic or nuanced things that go overlooked by the non-rationalist masses in favor of some matter of current events or personal interest.  All the same, all truths about artificial intelligence hinge on logic, and even the least severe assumptions or errors regarding AI on someone's part are still irrational.

An AI, whether truly sentient or not, could write sentences exactly like a human would, but there is no single way a human can write a sentence.  A human could use dashes improperly or not use them at all.  A person could write incredibly ineffective strings of run-on sentences or write in grammatically perfect statements (though grammar is an arbitrary construct that could be revised to some other arbitrary, contrived system).  Or, someone could use any other alleged conventions of generative AI writing, such as "It's not A; it's B" and could have started using them long before something like ChatGPT prompted a cultural association of this phrasing with AI.  It simply does not follow from human or AI origins that writings will have any particular syntax or punctuation usage.

Moreover, a human could try to write in a manner fallaciously assumed (as if all assumptions are not epistemologically fallacious) to be the hallmark of AI generation because they like the writing style or simply want to toy with idiots—it would always be the fault of the one making the assumption for being stupid.  Avoiding the use of dashes to tread carefully because some other people are fools with a false belief, though, does not mean someone is irrational, as long as they do not believe that dashes must be from AI and only avoid using them in order to navigate, for instance, the workplace without accusations of inauthentic writing.

Here is yet another example of multiple non-rationalists leaping onto a current events bandwagon philosophy that is utterly untrue.  There is absolutely no logical connection between particular kinds of punctuation and human or AI authorship.  At most, there could be a happenstance or situational connection, but this is not one of strict logical necessity as some might perceive it to be.  The relationship is thus never inherent and cannot prove that a given sample of writing has been brought about by one or the other.  As much as some might believe otherwise or want it to be true, there is no such thing as absolute confirmation that a given writing sample originated from humans or software!

To provide another example, bullet points with repetitive, brief, or very simplistic content in them could come from generative AI.  Entire articles might be stuctured in this manner that does not flow very well or really touch on distinct ideas or facts about a subject.  But again, this does not necessitate that AI must be responsible in any specific case because it is logically possible for someone to write like this (it does not contradict logical axioms!) because it is/already was actually their personal style, because they intend to parody or mock supposedly giveaway components of AI writing (whether they believe this or not), or because they want to actively lead others (who would only be assuming anyway) to think AI is responsible.

People are often very incompetent both in grasping logic and in articulating themselves, so on this level as well, it would not matter how illogical what is being articulated is or how poorly or repetitively it is worded or punctuated.  Try as hard as you might, you will never know with absolute certainty—the only way to truly know something—from the writing output whether it was created by a person or by a generative AI.  You can know in the moment that you are writing something.  You can know it is logically possible for humans or software to produce writing of a given sort.  As for writing that is circulated on the internet or provided to you by another person, at best, you might have highly fallible evidences that never amount to logical proof.