Via a series of multi-panel comic strips^[1] that I found posted and reshared on social media an interesting question was posited: what if [insert great artist/scientist/thinker here] had used AI?

We take for granted that the greatest people of our collective creative human history never leaned into the modern thinking simulators to which we have access, but perhaps Einstein or Shakespeare would have pasted drafts of their ideas and work into a chatbot had they had access to one. Thus they might have buoyed their feelings of uncertainty and modified their work under the advice of an AI, y’know just like anyone else might do nowadays.

So, the gag set in motion by the aforementioned comic strip was simple: what if someone like, say, Darwin had asked AI about his nascent Theory of Evolution or Mary Shelly had queried her thoughts about her plot outline for Frankenstein^[2] past a chatbot? In doing so the cartoon version of the LLM had replied with expected kind of feedback.

Anyone who has done such a thing, asked an AI model for input, would immediately recognize the flavour of formulaic, pandering, sycophancy that generally emerges, and which was illustrated in the cartoon.

The likely scenario was illustrated: in each panel of the comic, the gag AI had showed reticence about straying too far from established notions, concepts, and norms and had instead discouraged the cartoon human from pursuing further work on their ideas. Darwin was discouraged from pursuing evolutionary theory^[3]. Shelly was nudged to stick with more commercially viable women’s fiction.

It was just supposed to be a funny cartoon.

…but like any good bit of humour the notion rang a little too true for me and highlighted a subtle issue that I think many people already sense about generative AI and find difficult to articulate: that is, our experiences with these systems has almost uniformly resulted in a kind of blanket conformity to existing patterns. It all results in a kind of ersatz creativity within the bounds of what is already known, a pattern matched against existing reality, and a simulation of creativity jumbled together from the pattern-spewing innards of an AI engine.

AI tends to shave the rough edges right clean, and all that is left is a boring, unoriginal, polished stone.

Creativity is hard to pin down, after all.

Creativity is something we, even as meat-brained humans, often struggle to define in a satisfying way.

We wave hands past a definition that includes even more vague terms like imagination and originality, overlooking the hard truths that we are ultimately products of complex biological outcomes and a lifetime of experience. AI approximates creativity in a believable way because, I think, it is difficult to pin down even in ourselves, let alone others. We buy the act. Yet, there is an invisible novelty that fails to manifest in these systems. And that novelty is what is important because we don’t know it is missing. We don’t miss what we don’t know exists and thus cannot see.

We may more and more feel the absence of bold ideas outside the metaphorical box as we more and more lean into these tools, but it is impossible to see what never was.

So, what if Darwin had used AI… and walked away from his theories because a chatbot had discouraged him with a list of reasons why those theories didn’t fit current thinking?

And what if Shelley had used AI… and wrote a novel more fitting to the styles, trends and commercial viability of the fiction styles and demands the day? Or, what if she wrote to the nineteenth century equivalent of the booktok algorithm?

The problem with creative brilliance is that it is unique and rarely fits the existing pattern. It breaks it, and we often don’t recognize it even when it comes from ourselves, and maybe especially then.

What if you used AI… and walked away from something interesting because you dreamed outside the establishment when AI pushed back?