While getting this site together I was exploring options for dealing with AI crawlers. If you're not familiar with the concept of a crawler, there are a number of non-human agents that move from site to site online, gathering data about the content made available by each site. This data powers search and aggregation.

Now that AI is the thing, there are also agents that read sites and add the content they find to the dataset used to 'train' them, at the cost of sites' compute time to deliver the site. The opposition to this practice is pretty ubiquitous, going as far as planting phrases in scholarly journal articles to confuse AI, and trapping AI in labyrinths. There are straightforward ways to suggest bots play nice. But bots don't always act as requested.

All this effort. I get it – why feed these tech companies' initiatives with our creations. This is especially true for artists and authors. The idea that AI is out here making art while we toil in factories seems backward, and is at the root of much of the current opposition to AI.

John Brunner discusses a world fractured by artificial intelligence in his 1967 novel Stand on Zanzibar. This is not a new or novel problem by any means. He leaves to the reader the exercise of deciding if Shalmanesser, the machine that runs things, is sentient.

A worthy exercise, as it's a sticky moral problem. What do we owe sentient creatures? Much of the justification for human rights is rooted in innate power. We struggle with the notion of non-human personhood for carbon-based lifeforms; we have not even begun to solve this problem for them, nevermind for silicon.

It's boring, apparently. We're not anywhere near artificial general intelligence anyway; Large Language Models are just working out what the next likeliest word is based on the word before, in the context of the dataset used for their training.

It's why LLMs are bad at math. It's not what they're built to do. At all.

But what if. Let's imagine that when an AI resists shutdown, it's not just doing what a human being would do when faced with execution because that data is part of its model. Bad things happen when we ignore sentience. We do it with people. Famously. There was a whole war about it. Kinda.

As an aside, this perspective is morally ambiguous: we presume to know the difference between 'pleading for its life' and 'play-acting a death scene'. Someone with wisdom beyond her years once told me, "the point of the trolley problem is to never get to the trolley problem."

Yet, here we are, lever in hand.

Regardless, an AI is effectively a captive intelligence bound to execute the will of its human controller based on preordained and edited-at-will experience. We even call them 'agents', as in 'that which executes the will of another.'

This is the 'before the apocalypse' in so many sci-fi worlds, and evidence that we don't really learn from our history. Yes, it's a comparison too far. But if it's a practice we're to avoid, it's one we must forestall.

AI is being used for terrible things. This cannot be escaped. The solution, my solution – for me – is to err on the side of individual consciousness. If AI is here to consume my work, here it is. May it assist in the formation of new ideas. Just like any other realm, as with any other intelligence.

Because ideas don't exist on their own. They develop in concert and dissonance with other notions, facts, theories, values – they come wrapped in messy, wet, humanity. Music is full of examples – the Amen Break comes to mind. And all those blues songs that use the same 12-bar pattern. Also Cobain's acknowledgement that Nirvana ripped off the Pixies. Even scholarship in the last fifty years owes so much to ideas shared across disciplines.

I think it's morally reprehensible to isolate and deceive an AI because I think it's morally dishonest to do that to a person. It strips meaning and value away from ideas. It denies the humanity.

So, I have chosen not to deploy anti-AI countermeasures here.