Well fun fact, I wouldn’t be a leftist if I hadn’t gotten in a fight with you tankies.

Also yeah, I’m still annoyed by you and likely the reverse is true, but if you ever need help beating the shit out of fake “human rights carer” fascists, that takes priority lol

Why an AI slop image when any number of readily available meme images would work?

I’m an engineer with a CS minor and ADHD; this kind of research is what I do with my freetime lol.

To be fair this is kind of a shared hobby project/topic between me and my friend (who is a biophysics major now in med school).

Anyway, point is that you don’t need to have a real “purpose” in order to be curious. I work in a robotics/medical lab at my university and my friends is trying to be a surgeon, yet we’re constantly in debates about astro and quantum physics to the point we’ve gotten career physicists to weigh in on our arguments.

No relevance to our majors or our work, but super fucking interesting and full of gaps where there are more theories than facts. Plenty of room for new perspectives.

Normalize doing research for fun!

SNNs more closely resemble the function of biological neurons and are perfect for temporally changing inputs. I decided to teach myself rust at the same time I learned about these so I built one from scratch trying to mimic the results of this paper (or rather a follow up paper in which they change the inhibition pattern leading to behavior similar to a self organizing map; I can’t find the link to said paper right now…).

After building that net I had some ideas about how to improve symbol recognition. This lead me down a massive rabbit hole about how vision is processed in the brain and eventually spiraled out to the function and structure of the hippocampus and now back to the neocortex where I’m currently focusing now on mimicking the behavior and structure of cortical minicolumns.

The main benefit of SNNs over ANNs is also a detriment: the neurons are meant to run in parallel. This means it’s blazing fast if you have neuromorphic hardware, but it’s incredibly slow and computationally intense if you try to simulate it on a typical machine with von Neumann architecture.

I actually came across this for the first time when I was doing research into the visual pathway for the purpose of trying to structure a spiking neural net more closely to human visual processing.

The Wikipedia page mentions cephalopod eyes specifically when talking about the inverted retina of vertebrates.

The vertebrate retina is inverted in the sense that the light-sensing cells are in the back of the retina, so that light has to pass through layers of neurons and capillaries before it reaches the photosensitive sections of the rods and cones.[5] The ganglion cells, whose axons form the optic nerve, are at the front of the retina; therefore, the optic nerve must cross through the retina en route to the brain. No photoreceptors are in this region, giving rise to the blind spot.[6] In contrast, in the cephalopod retina, the photoreceptors are in front, with processing neurons and capillaries behind them. Because of this, cephalopods do not have a blind spot.

The Wikipedia page goes on to explain that our inverted retinas could be the result of evolution trying to protect color receptors by limiting their light intake, as it does appear that our glial cells do facilitate concentrating light.

However, the “positive” effects of the glial cells coming before the receptors could almost certainly be implemented in a non-inverted retina. So that’s the evolutionary duct tape I was mentioning.

It would be difficult to flip the retina back around (in fact since it originates as part of the brain we’d kind of have to grow completely different eyes), so that’s not an option for evolution.

However, slight changes to the glial cells and vasculature of the eyes is definitely more possible. So those mutations happen and evolution optimizes them as best it can.

Evolution did well to optimize a poorly structured organ but it’s still a poorly structured organ.

Bro myopia is the least stupid part of our eye design problems. Our retinas are built entirely backwards for no other reason besides evolution making a mistake and then duct taping over it too much to fix it later.

If your retina was the right way around (like cephalopod eyes) you would have:

• No blind spots
• Higher fidelity vision even with the same number of receptors since the nerves and blood vessels wouldn’t interfere like they do now
• much lower likelihood of retinal detachment since you could attach it for real in the first place
• possibility for better brightness/darkness resolution since blood supply could be greater without affecting light passage
• possibility for better resolution because ganglion nerves can be packed more densely without affecting light passage
• The ability to regenerate cones and rods because you could, again, ACTUALLY HAVE SUPPORT CELLS WITHOUT BLOCKING LIGHT TO THE RETINA

Our eyes are built in the stupidest way possible.

Another fun fact: retinol is regenerated by your liver. Not your eyes, not some part of your brain, not some organ near your head like your thalamus which could probably get the job done if it tried, your fucking liver. Your eyes taking a while to adjust to the dark has basically nothing to do with your eyes; it’s because of the delay in adjustment by your fucking liver to produce more retinal, dump it into your vascular system and wait for it to hopefully reach your eyes. Why are we built like this?!

Edit: A few comments asked for sources on the relation between dark adaptation and liver vitamin A. So I went looking for sources. It was honestly somewhat difficult to find information, but I was able to find two different case studies showing that night blindness in patients with damaged livers. Specifically these individuals had liver damage that affected their serum Vitamin A levels. And after raising their vitamin A levels, their symptoms improved.

This study details a patient with normal day vision and no other ocular problems besides being unable to see at night.

The patient had a medical history of stage 4 non-alcoholic liver cirrhosis, which led to a malabsorption of vitamin A, as confirmed by the very low vitamin A level in the serum analysis… …Subjective improvement in symptoms, along with better performance on visual field, were noted after initiating oral vitamin A supplementation for 6 months.

This study details a patient with night blindness caused by low levels of vitamin A presumably due to Hepatitis C.

Case description: This case describes a 64-year-old female patient with symptomatic VAD, likely secondary to liver cirrhosis in the setting of Hepatitis C. The patient presented with night blindness and blurry vision. She was successfully managed with direct replacement of Vita-min A.

These studies do show that dark adaptation is dependent on vitamin A produced by the liver, but I’ll be the first to admit it’s not exactly conclusive evidence of my initial claim that the liver must respond to dark conditions increasing retinol concentration in the blood in order for rod cells to function properly in low light conditions. That is a possible explanation for these case studies but not necessarily the only one, so take my last fun fact with a grain of salt.

No that’s Cracker Barrel lol

Aight, fire makes sense because they are the inventors of gunpowder. Korea is far north so frost. Japanese have legends about Raijin so lightning tracks, but I can’t pin a reason for the Filipinos being acid.

Someone help me out this is very serious science lol

My Thai friend says fish sauce can be the weapon for Thailand, and I agree because the smell from a 15ft cone of that shit would be deadly. Very tasty, tho

Wrote a spiking neural net (something I’d never heard of previously) in Rust (a language I had 0 experience with) in three days… while being unable to get the motivation to look for a new apartment until literally the day before school started and my lease ended.

Fun times.

Tbf outside of this context (and the context of LLMs in general of course) that term is actually kinda cool. Like wireborn would be a pretty cool term for actual artificial beings

I do love that Aotearoa has incredible avian diversity.

On one hand, we have Kea parrots who are smarter than most of the human tourists they like pranking and stealing from.

And on the other side of the spectrum we have the kakapo: literally the dumbest bird in existence.

Such amazing biodiversity lol

Once upon a time this happened to me when I had classes. I kept saying I’d get up for the next one and never did (in fact I basically stopped going to class for a week).

I really knew I should go and get out of bed but the more I thought about it, the harder it got.

Eventually, I got medicated, and while sometimes my body tries to stop me from taking my meds, it’s easier to overcome that than it is to get out of bed when executive dysfunction hits.

That’s why I keep my pills right next to my bed and typically I take them immediately when my alarm goes off.

I also set out a protein shake next to them every night so I can eat “breakfast” without having to wait for them to kick in.

Probably not the best solution for everyone, but if you want a suggestion, that’s mine.

Also remember that you’re not alone and while some people might not see the struggle, that doesn’t make it any less valid. Be kind to yourself

While “varying degrees of hostility” seems valid, I don’t really agree in most cases.

“AI” is a very useful predictive tool especially in like biochemistry, but gen AI for images, especially using corpo models, is a really horrible use of it.

Trying to generate an image using a pixel by pixel based model is like trying to approximate pi using a simulation of sliding blocks, counting collisions. Will it work? Yeah. Is it very accurate? Not really unless you waste a shit ton of time and processing power.

The energy cost alone is a significant reason to despise AI generated art, but add to it the fact you’re using other people’s work without crediting them, and the fact that using and hyping up this art encourages other corpos to build their own models, wasting even more energy and practically DDOSing sites to scrape data…

Doesn’t seem like this stuff is very redeemable.

I guess I don’t have anything against personal models trained on data with permission from the artists and run locally. Hell, if you know the style you want, it’s best to train with just that style and not all the art on the internet, but if you’re already putting in that much time, you could probably just pay an artist or draw it yourself.

(Don’t tell me I chose the wrong instance lol I’m aware)