Ethical considerations aside, where I have long found myself fundamentally against the disrespectful and extractive ways in which AI has been developed and its models trained, I think the line I draw when it comes to accepting the use of AI or the product thereof is at cognition, intention, and humanity. For instance I cringe mightily when AI is used to slop out videos on spiritual subjects, a matter that is uniquely human in nature. I cringe only slightly less when it is used in a manner where it is clear that a human had no role in guiding its output beyond the prompt or automation harness.
I can accept AI’s use as a tool that is subservient and supplemental to human cognition, not replicative of it or as a replacement for it. A supplement rather than a stand-in. If it’s a one-and-done set-and-forget implementation of AI, that’s where I start seeing the telltale marks of AI, and that’s where I start intuiting a fundamental laziness in the user. AI output is best used as an input in an overall process, not as the end product; something to act as an interlocutor or a subsequently and independently verified information source or processing medium, not as a replacement for the human cognition or creativity that was the initiating reason for the process to begin with.
AI output should ALWAYS be questioned and interrogated by a human, or else the human who relies upon it too much risks atrophying their critical thinking skills and their very intellect. This much has been borne out in a growing number of studies on the matter, after all.
I thus personally use AI only sparingly, as an interlocutor to bounce thoughts and ideas off of, and I carefully examine every response for sycophancy and hallucination. Even then, only because my employer encourages (tacitly requires) it. It is of paramount importance to me that I maintain my mental faculties, critical thinking skills, and subject matter expertise (probably in no small part because from a young age I made intelligence my core identity trait in response to persistent bullying and ostracism in physical sporting pursuits).
I once came close to the edge of AI psychosis when I used it too heavily for relationship advice at a time when I was struggling with stressors and my degrading responses to them as they built up over time. I would blame that and my flirtation with alcoholism for the end of that same relationship. The AI chatbot was too eager to validate my misperceptions and accept my emotionally-loaded impressions as absolute fact and then advise appropriately. And how could it not make such mistakes? It has no life experience that it could refer to in order to intuit that I could have been out of sorts, let alone in the midst of the process of becoming inebriated.
It is precisely because AI is, despite having ingested and processed a huge amount of human writing, fundamentally untethered from the human experience that it can only ever be a supplement to human judgment and cognition, not a replacement. And even as a supplement, one must always keep certain rules in mind when working with AI:
- AI will say whatever it can to please you unless challenged repeatedly and assertively.
- AI will not understand the context or the bigger picture of the situation unless you tell it, so tell it why you need the answer and what you intend to do with it.
- AI will often misquote or misinterpret things for various reasons without telling you. For instance, I asked it for a monologue from a particular play for an audition, and it presented a paraphrase from that play and claimed it was from that play: in fact, the text was not present in that play at all. When pressed, the AI admitted that it could not reproduce the text for copyright reasons.
- AI will not be forthcoming about things it does behind the scenes based on rules over which you do not have control unless pressed and questioned.
- AI does not know the current date nor does it have an intuitive sense of time like you do. It cannot perceive the ambient light levels of the time of day like you can. You must be explicit every time you pick up a conversation with what date and time of day it is if you want it to be able to “comprehend” the passage of time.
- AI is not human. AI is not embodied. AI is not conscious, at least not in the way humans are. It has nothing that either theists or atheists would consider a soul or “that which makes us human”. Consider that in your interactions.
One last thought on a more ethical front: I have seen a growing trend of local businesses imploring local residents to support them in the name of “buying Canadian”, yet opting to use American-owned AI platforms to generate advertising copy and content rather than supporting the local businesses that provide writing and graphic design. I believe this is deeply hypocritical and an incredibly shameful and deceptive practice.
If you use AI, like I am regularly encouraged to do as a part of my day job, do it ethically and do it in a way that aligns with your moral values and preserves your agency and your intellect. Because if the AI bubble bursts but you’ve let your mind go, you will have nothing.
This content was written without the assistance of LLM-based generative AI.