If there’s one definitive act of foolish accommodation that I have come to regret over the past few years, it’s the hollow concessions that I found myself making with my students over the use of AI. My standard position is that I don’t want my students to use AI at any point in the creative process. Some have asked me whether they could use AI to tidy up their grammar and formatting after they’ve written their drafts. I tell them that I would rather see those mistakes so that I can help to correct them with my feedback. I also point out that they could set off the AI detection software when submitting if they used it at the last hurdle of their assignment. This also heads off the standard level of defence when a student submits work that I strongly suspect was made with the help of an LLM. They tend to say that they might have used it in a later editorial phase –⁠ which is also the standard form of mea culpa some writers have used when their novels turn out to have been written with the assistance of an LLM. But still, when students have asked me about acceptable use of AI, I have tried to come up with some kind of compromise, a little breadcrumb to throw their way.

This weakness has perhaps resulted in the worst advice I've given to my students about the acceptable use of AI. I told them that they could use if for brainstorming and then write the poem themselves once they find an idea that clicks. It might have been that I was once again being the diplomat, looking for that one exception that makes me feel less obstinate and unaccommodating.

The reason I think that this advice is bad, in the sense that I think it does more harm than good, is that I was (in effect) instructing my students to give up on their own imaginations. I was also falling for the falsehood that the imagination is only there for the brainstorming stage, and then absent for the part where the craft happens. But the imagination is there at every point of the process, even in the edit, where we try to reignite the original vision from our initial textual rendering to see if we got close to that initial spark. Telling students of creative writing that it's okay not to use their imagination at one of the most vital points not only compromises their natural capacity for inception, but also weakens their ability to rely on their imaginations as the project progresses.

(This also says more about my own prejudices and assumptions about authorship, that the ideas spring from the ether and are not really ours before we make them our own by exerting our own intentions and effort upon them. Underneath all this are the usual assumptions about effort and labour that I have been hypnotised into thinking are the only signifiers of authenticity.)

This isn't just about one's ability to create. Our imaginations are an integral part of how we are able to live, as well as what gives our lives value. We dream every night, but we also dream throughout our days. Over the first few decades of our lives, we are taught to navigate the practical and social elements of everyday life but we aren't necessarily trained to use our imaginations (or maybe we are by the things we watch, listen and read –⁠ for better or worse).

When we speak of a human being or a cultural artefact as “soulless”, we often mean it in the context of there being a lack of imagination. The classic philosophical zombie has nothing going on inside but is still able to function and is outwardly indistinguishable from a conscious human being. Many philosophers say that the P-Zombie is logically impossible, that if it exhibits all the outward traits of consciousness then it must be fully conscious. I bring this up because I'm about to say something quite unreasonable and possibly offensive. In my everyday life, I have come into contact with people who are so dependent on AI that they are unable to do many things without it, from their emails and text messages all the way up to their critical and creative expression. All elements of their creative and knowledge work has become modulated by it. AI has placed itself between them and the world at every available point, and I've been trying to get a handle on a certain aspect of their characters that unnerved me whenever I made contact with them. And that impression that they gave was of being soulless.

I'm not saying that they were P-Zombies, more that there seems to be a barrier between them and the world, as if they couldn't express themselves or receive the expression of others without running something by the chatbot first. In doing so, they had already surrendered their imaginative, emotional and rational faculties to the machines. Emotion, imagination, reason –⁠ aren’t these the terms we most often reach for when asked about the qualities that make us human?

This is the point where I have to perform my usual disclaimers and confessions. I still use LLMs, normally for boring stuff like admin and my accounts as well as vibe coding elements of my digital garden. I have also, due to the unreliable nature of my work and employment, trained AIs as part of a side hustle that utilises (rather than employs) a growing, precarious resource pool of humans that are paid per task to check AI output for accuracy and fidelity.

When speaking to my eldest about how I used AI to save myself a couple of days reading up on and experimenting with how to create something simple on one of my websites, my brilliant progeny still sassed me with, "But Dad, you didn't learn anything?"

They're right. Every time we choose to use AI, we sacrifice something of our humanity. Not only do I lose out on exerting my brain and pushing myself to achieve something in a field that is quite contrary to my skillset –⁠ I also miss out on the little sense of triumph that erupts when the thing that has been failing all week suddenly works.

Even then, the LLM is always making little suggestions and offering to help out with other stuff, like a blurb for an intro page for the site I'm building. I repeatedly tell it that I never want it to write anything for me. It complements me on my integrity before making another suggestion a few posts later.

A lot is made of the more distinctive perils of AI, from the environmental carnage, the rocketing prices of consumer electronics due to the current RAM shortage to the current car-crash journalism on AI psychosis. But not as much is made of the quiet dilution of our own humanity when we come to overrely on this technology.

It could be that we have already been diminished by other digital forms, such as the flattening effect of algorithmic social media, how the vast array of our reactions have been whittled down to likes and dislikes. In this sense, it’s interesting to see that same governments and educators that have decried the effect of social media on our young are also gung ho about advocating for AI in education. With AI writing our emails, being our counsellor and companion, and offering to take the wheel at every stage of a creative project, I see a further flattening, a dilution of our humanity, but applied at barely noticeable degrees at every available element of human consciousness, expression and experience.

Wallace Stevens once spoke about poetry (and by extension, the imagination) as being, “a violence from within that protects us from a violence without.” I’ve always found this particular soundbite to be overly dramatic. But I do believe that the imagination is a necessary faculty when ideas –⁠ such as AI’s dominance being inevitable and that if we don’t get on the train then our competitors will get on it and leave us in the dust –⁠ are so readily accepted and parroted by our own governments. I trust my imagination much more that I trust the visions and epiphanies of the creatively and spiritually bankrupt. And I tend not to get onto trains if I haven’t been informed about their destination.