AI Blog Question Challenge

Rishabh P. Sharma proposed a blog challenge about AI, and I found the questions thoughtful enough that I wanted to give it a shot. Here are my answers!

How was your first experience with AI models?

I tried ChatGPT once for suggesting character names that fit within a particular theme or vibe. It seemed more useful than it ultimately was; lots of things slipped in that were not really appropriate answers, and I’m unconvinced it was any better than just browsing a baby names book or website.

Do you use AI or are you completely against using it?

It seems impossible to completely avoid it, with numerous services outright replacing their help documentation and website searches with chat bots. Therefore, I avoid using it as much as possible, and I’m generally very opposed to the proliferation of the current crop of LLMs, which were trained on stolen work, encode status quo political and cultural biases, and whose sprawling data center requirements are a rolling environmental disaster.

Do you have any preference among different models, for example Claude vs ChatGPT? If yes, how do you choose?

I don’t care about functionality or correctness if the model is trained on stolen work, contributing to boiling the planet, and likely tanking the economy when the current AI bubble bursts. All such models should be destroyed.

If a model’s training corpus can be proven ethical, and if its performance can be optimized to not require massive fossil fuel burning data centers, and if its economics can be safely constrained, then I might start to be more interested, but even then, only for very narrowly defined use cases. AI is a fancy statistical search, not an omniscient machine god, and there are only just so many tasks for which a fancy statistical search is the right tool.

What aspect of AI models do you like and what do you not like?

I like that they can take an overwhelming amount of complex and inconsistent data and essentially run natural language queries against it.

I dislike that so many people think that’s a good way to generate art, that they’re trusting a pile of linear algebra to write business critical code and make public policy, and that they think it’s a conscious machine god that will ultimately merge with or outright replace humanity. This is TESCREAL cult shit, and I am terminally exhausted by it.

How do you feel about AI generated images? Does it annoy you if someone use them in a blog post?

In my game development background, I was (and am) fascinated by procedural generation, most deeply explored in the roguelike genre. AI image generation isn’t the same thing, but you can argue they’re cousins of a sort: either way, you’re using algorithms to create art.

That said, in (good) roguelikes the goal of proc-gen is to build responsive gameplay constraints, not to generate the whole game. The proc-gen algorithm is itself a work of art, carefully tuned—by a human!—to emphasize certain types of outcomes and have a particular feel. Today’s AI image generators, by contrast, are just brute-force smashing together everything they could scrape off the internet and producing a bland little-bit-of-everything style that lacks originality and human intent.

When you’ve built and tuned your own model that intentionally expresses your own human creativity in its patterns and parameters, then we can have a conversation.

Internet is flooded with AI slop now, full of generated text, images, audio and videos. How do you filter it from authentic human creation? Do you have a strategy?

Search engines’ “no AI” filters worked way better a few months ago than they do now, and people are currently incentivized to hide the fact that their work is AI due to (correct) fears of backlash, which makes filtering it out nearly impossible, unfortunately.

I’m really worried about how much this is all becoming a vibes check, which is already leading to witch hunts over false positives. And I’ve seen anecdotes like students insisting real Renaissance artwork must be AI because “humans can’t do that”, which is such a tragic surrender of human potential.

The best answer I have here is to connect with real people in the real world as often as possible: go to art shows and conventions and indie markets and community groups and maker spaces and revel in each other’s mutual and unique existence.

Are you hopeful for a better future with A.I. or a dystopian one?

I’m more of an AI doomer, mostly because it’s being employed to shore up the political status quo—late-stage capitalism, oligarchy, and plutocracy—which is itself unsustainable. The tech bros run their mouths about a post-labor society, but history has shown they’ll just hoover up all the wealth and wall themselves off in their private compounds and leave the rest of us to fend for ourselves, all while blaming us for the very conditions they forced us into.