Hacker News Reader

Comment by matsemann | Hacker News Reader

On one hand, it's great that more people are able to more easily make things like these. They're a great tool for understanding, but obviously an enormous amount of work.

But on the other hand, I also wonder if those that get pumped out now, also uses up all the oxygen in the room for the really great ones.

Like, someone shoddily slaps together one for concept X. It's not great, has some misunderstandings, and what it visualizes is mostly random and not the vision of the maker. But it goes semi-viral, and beat everyone else to the punch. So it ruins someone else's motivation to do it properly.

View on HN

I'm not sure it ruins the motivation. e.g. Rayleigh scattering is one of those norotiously bad explanations that's for some reason still bad. At some point I ended up on the topic with my 5 year old the other day, so I figured I'd check what's out there, and yeah, it's still bad: everything starts with "why is the sky blue" and talks about scattering which doesn't mean anything when you're 5. So I have codex whip up an animated demo for what the question should be: why isn't the sky black? Once you understand that that's the correct question, then you can properly add in light bouncing in the air as an effect, and that blue bounces more, therefore you see blue at off-axis angles (the whole sky), and red remains in the direct rays (c.f. red sunsets where most blue has scattered away from the longer rays).

Anyway, the motivation here is just that the existing explanations are already bad, and AI helps with motivation here because it makes it super easy to do a better explanation if you know what you want. I doubt that the it really sunk in and that she could properly wrap her head around it yet, but I'm sure that whenever she's ready to really think about that, it's a much better explanation than any of the ones that exist today.

> Rayleigh scattering is one of those norotiously bad explanations that's for some reason still bad. At some point I ended up on the topic with my 5 year old

I am in awe of your Daddying, whether nature or nurture.

You should reproduce more!!!

What really impresses me about these AI visualisations is that they're often (but by no means always) fully screen reader accessible, with proper keyboard and ARIA support. Those by Ciechanowski very much aren't, and even vibe-coded slop accessibility is better than no accessibility at all.

Accessibility with AI is hit-and-miss; if you don't tell it to make things accessible, it either will or it won't, and then it'll just continue with whatever trajectory it chose. If you tell it to do accessibility initially, it'll see its own ARIA later and continue in a similar vein.

From very anecdotal experience, I find OpenAI models more likely to consider accessibility for greenfield projects, even when not asked to, but this might just be an accident of statistics.

Reply on news.ycombinator.com