Re: That Post Staff Has Always Been Awful To Trans Women But I’ve Been Here About Over A Decade And

All fancy smancy generative ai models know how to do is parrot what they’ve been exposed to.

A parrot can shout words that kind of make sense given context but a parrot doesn’t really understand the gravity of what it’s saying. All the parrot knows is that when it says something in response to certain phrases it usually gets rewarded with attention/food.

What a parrot says is sometimes kinda sorta correct/sometimes fits the conversation of humans around it eerily well but the parrot doesn’t always perfectly read the room and might curse around a child for instance if it usually curses around its adult owners without facing any punishment. Since the parrot doesn’t understand the complexities of how we don’t curse around young people due to societal norms, the parrot might mess that up/handle the situation of being around a child incorrectly.

Similarly AI lacks understanding of what it’s saying/creating. All it knows is that when it arranged pixels or words in a certain way after being given some input it usually gets rewarded/gets to survive and so continues to get the sequence of words/pixels following a prompt correct enough to imitate people convincingly (or that poorly performing version of itself gets replaced with another version of itself which is more convincing).

I argue that a key aspect of consciousness is understanding the gravity and context of what you are saying — having a reason that you’re saying or doing what you are doing more than “I get rewarded when I say/do this.” Yes AI can parrot an explanation of its thought process (eli5 prompting etc) but it’s just mimicking how people explain their thought process. It’s surface level remixing of human expression without understanding the deeper context of what it’s doing.

I do have some untested ideas as to why its understanding is only surface level but this is pure hypothesis on my part. In essence I believe humans are really good at extrapolating across scales of knowledge. We can understand some topics in great depth while understanding others similarly on a surface level and go anywhere in between those extremes. I hypothesize we are good at that because our brains have fractal structure to them that allows us to have different levels of understanding and look at some stuff at a very microscopic level while still considering the bigger picture and while fitting that microscopic knowledge into our larger zoomed out understanding.

I know that neural networks aren’t fractal (self-similar across various scales) and can’t be by design of how they learn/how data is passed through them. I hypothesize that makes them only understand the scale at which they were trained. For LLM’s/GAN’s of today that usually means a high level overview of a lot of various fields without really knowing the finer grain intricacies all that well (see how LLM’s make up believable sounding but completely fabricated quotes for long writing or how GAN’s mess up hands and text once you zoom in a little bit.

There is definitely more research I want to do into understanding AI and more generally how networks which approximate fractals relate to intellegence/other stuff like quantum physics, sociology, astrophysics, psychology, neuroscience, how math breaks sometimes etc.

That fractal stuff aside, this mental model of generative AI being glorified parrots has helped me understand how AI can seem correct on first glance/zoomed out yet completely fumble on the details. My hope is that this can help others understand AI’s limits better and therefore avoid putting too much trust into to where AI starts to have the opportunity to mess up serious stuff.

Think of the parrot cursing around children without understanding what it’s doing or why it’s wrong to say those words around that particular audience.

In conclusion, I want us to awkwardly and endearingly laugh at the AIs which mimic the squaks of humans rather than take what it says as gospel or as truth.

Background decorative elements

I felt like the placement of your background images within the body tag to the top and bottom of your main content div is pretty intuitive overall. The only thing I could think of to improve the vines would be to make them pseudo elements (another YouTube video by Kevin Powell) of the body element so that way you don't have to have empty divs cluttering your html for them. Pseudo elements are useful for purely decorative elements because they're created purely out of CSS; no HTML required!

Now for color: below the fold bc hyperfixation go brrr

TLDR: Color contrast is hard; especially with the current color systems we have today (HSB/RGB) since the numeric incantations of popular color systems don't line up with how our eyes actually perceive color, especially when it comes to brightness. Google made a new color system (along with a tool based on that color system to generate color palettes) that lines up better with how humans perceive color but it still isn’t perfect. IMO using tools like palettte.app (with 3 t’s) and hand-selecting colors numbered by how light they are - more about that below - works best and gives projects more of a hand-made feel but Google's tool works well enough for making a quick and professional looking color palette.

Now for a very colorful rant:

Making CSS variables with the format

colorName-brightnessValue

makes it easier to meet contrast standards when coding up a website. I'll usually organize my colors to range from 0 to 1000 exclusive where 0 is black and 1000 is white. I usually use 100, 300, 500, 700, and 900 the most so I have those 5 shades of each main color ready to go whenever I start a new project. (I use the same 4 colors for all my projects so it's easy to copy and paste the colors from the last project into the new one!)

That way if I know my background is violet-100 and my text color is violet-700 that the text will be legible because of the difference in brightness values is 600 just based on the variable names alone. Generally a difference of at least 500 (assuming you go linearly in perceived brightness) is enough to get AAA contrast levels.

For an example, here are my css variables for zephiris.me:

Others do the numbering of the shades differently. Here's how tailwind does it:

Tailwind CSS color guide's section on making a custom color palette has links to some helpful tools on how to hand-pick several shades of a set of colors so that the brightnesses of the colors decrease with the numeric value in a way that actually lines up with people's perception of color.

On that same page, tailwind also has all those color palettes shown above free for you to steal use in your projects!

Google's Material 3 design framework has shades of every color go from 0 to 100 for brightness instead of my 0-1000 brightness or Tailwind's 0-1000 darkness.

Google also has a nice resource which will automatically generate a color palette for you (I like doing it manually with palettte.app [with 3 t's] but if you just need a quick and easy palette that meets contrast standards, Google’s resource works pretty well).

Google actually has a very interesting blog post on why it’s so hard to make accessible colors and get the right amount of contrast with current color systems like HSB/RGB etc.. To fix this, they created a new color system, HCT (Hue Chroma Tone), which helps solve that problem by having HCT's Tone value better match our human perception of brightness.

Wrangling Several Colors to Work Together

My main advice for overall UI design is to pick three colors: a primary color (for background colors), a secondary color (for card backgrounds/text colors), and a tertiary color (for any elements that should be interacted with like buttons and links). I made a site to try out various combinations of colors and share those palettes with others a while ago (apparently November of 2021, thanks GitHub!)

Recently Juxtopposed made a more professional version of what I made called realtimecolors.com along with an accompanying video. Her website also features a palette generator where you just enter in one color and it will generate a set of palettes off that one color!

Unfortunately neither mine nor Juxtopposed's websites support multiple shades of each key color, so to get those additional shades make sure to either use Google's generator or one of the tools mentioned by Tailwind once you have a general set of colors that you feel works well together.

Color Palette Inspiration

What works best for me is to pick colors based on environments which I enjoy to be surrounded by. I based my color palette for zephiris.me on:

the night sky's bluish-purple hue (maybe I wear rose-tinted glasses ok?)

greenish-blue seafoam from ocean waves

golden rays of sun filtering through pine trees

Lastly, I used the trans flag colors to describe my gender for obvious symbolic reasons - I also like being next to a particular shark :p

Conclusion

Meeting color contrast standards can get way easier by numbering your colors based on how light they are. There are plenty of ways to get a set of colors labeled by lightness:

using palettes already made from Tailwind

Have Google do some math wizardry to generate you a custom palette

Use tools like palettte.app to create your own set of colors to play with

Regardless of what option you choose, the overall added structure of numbering colors' brightnesses makes it dramatically easier to make incredibly legible, accessible, and colorful designs.

Colorful rant over!

i like your website! it looks very nice

especially the gradient colored text!! you used a separate font to make it more legible

whenever i try to do something like that, it always becomes really hard to read... maybe i should learn some basic web design?

my website looks like this and it took two days of fiddling with css

Thank you! The biggest thing with making text legible is making sure there is enough contrast between the text color and the background or make the text big enough that it’s legible even if there isn’t that much contrast. The best guide on color contrast that I know of is the Mozilla docs! If you scroll down to the solution part there it has many tools to check text contrast.

Since your website has a warm bright canvas background darker colors and gradients would work better and end up being more legible.

If you’re looking to learn more about web development and especially CSS I strongly recommend Kevin Powell on YouTube! His videos on flexbox and grid are very helpful in understanding those new browser features and making responsive websites (websites that look great on any screen size). For example, I used grid for the nickname table and for my projects so that on desktop those elements would be wider and shorter while on mobile they’d get narrower and taller.

I love your site too, especially the canvas theme with the green branch/orange leaves and the clever span box to show your favorite color complete with a title tag featuring the hex code!

To be clear my site took me at least 20 hours of fiddling and development to make. Feel free to look at the site code (and my commit history) on GitHub!

"boys will be boys" is not true. boys will be girls

made this tiny edit to this work by @​mxsparks@oulipo.social, so now it's a very convoluted dual language pun

guess this is an art blog now /j

& a bonus acoustic version of bless ur heart which never fails to get me to cry when I feel like I need to cry

like black people are present in every single fucking genre and scene and popularized and straight up created several but people are so fucking hell bent on finding every possible excuse to not engage with their music because its easier than trying to confront their own racism. like okay well if its truly just a disinterest in most rap music then surely you listen to black artists in other genres right? who am i fucking kidding. of course you dont.

they should let me enter a server room i would lose my fucking marbles

pleeeeeeaaaaaaasse pretty please let me in2 your server rooms i promise i wont fall in love with the machinery pleasepleasepleaaaaaase <-LYING

the hadron collider is like an angel to me

Lyrics that promised us the riches of heaven were written by the engineers of our own public hell.

-Sophiline Cheam Shapiro

From her essay, “Songs My Enemies Taught Me”

Sopheline Cheam Shapiro was 8 years old when the Khmer Rouge invaded Phnom Penh and took power in Cambodia. She is a classically trained Cambodian dancer and above she was writing about the propagandist lyrics sung in Cambodia, lyrics perpetuated by the Khmer Rouge. Those lyrics promised growth and prosperity and those promises turned out to be empty for Shapiro.

If you want to read more, PBS have interviewed and written about her: