Aug 9th - Tiger Studies

I'm going to walk you through my animal study process! This only applies to animals. I'm hoping to study cross-hatching (since I'm VERY bad at it) at some point soon, but in this case I very badly messed up the anatomy of a tiger and set out to fix it.
Tigers are a bit different from most big cats. They have a somewhat standout-unique face shape, a lot of fur which hides their anatomy, and a lot of patterns that further obscure their anatomy. These stripes evolved to cause visual confusion, breaking up their silhouette so the prey don't see them until it's too late.


The original drawing wasn't terrible, but the eyes were placed wrong and were too large, the snout was too short, and the whole angle of the head had been tilted upwards. This erased a lot of the menace of the original picture. One of the most important things about drawing an animal like a tiger is capturing the FEELING of it. Through history, tigers have been depicted as demons, kings, spirits of the forest, and more. In all cases, they're incredibly powerful and imposing. This tracks, since they're the largest and most dangerous big cat. They're also the only species of big cats to actively hunt humans instead of as an incidental or unique behavior. In comparison to lions, it's like comparing grizzly bears to polar bears.

Here we start on our first study page!

When I'm doing a study like this, I like to do a minimum of four poses. One at rest, one in motion, one "odd in motion" drawing, and a wild card. The wild card is usually what I'm struggling the most with. In this case, it was the face of the tiger in an odd pose. In the first drawing (lower left) you can see that while I was acceptable for an at rest pose or neutral expression, I didn't understand the face very well still, and that showed itself in the final (top left) drawing. There weren't enough guide-lines to manage the proportions properly.

The second page was more for me to think things through and process the anatomy of the animal.

Full anatomy studies where I go into the bone and muscle structure is not something I do often, but I am willing do do when the first study page fails. I also wrote a little bit about my drawing process, and how I start with a single "anchor point" that is in the middle of the face and decides the facial angle. The body can shift around a lot too so it's hard to make a judgement on where to place the animal on that alone. This "anchor point" mentality also applies to the rest of the body- as once I'm done with the head, I move onto the neck, then the shoulders and front legs which decide how the animal is standing, then the torso (which can have a bend to create a more fluid pose or be stretched out in movement) and then the hips, back legs, and finally end with the tail. I then moved onto the face. One of the hardest parts about translating a skull to a full face is that the eyesocket tricks us into thinking the eyes are much larger. The large eye sockets on animals that process a lot of visual information (such as night vision, high fidelity/high distance vision, and high color vision like humans have) are explained by the large amount of nerves to translate the raw data into visual information our brains can percieve. I also noted that tigers have very large jaw muscles. This is also hidden somewhat by the skull, but is hinted to by the jaw bone and large occipital bone. The large protruding occiput is used to anchor neck muscles, which stabilizes the head during biting action. Jaguars have the largest bite force of all big cats, and their occiput sticks out further and is thicker than the one on a tiger skull. I also noted how the nose sticks out to make the profile of the face flat. Finally, I sketched out the full skeleton. This was mostly to inform posture.

The third page was to sketch the muscles to draw more information from the bone structure.

This is where I learned about their jaw muscles and large arm and shoulder muscles. I try to not reference off the art of others often for anything but style studies and then I never post it, but it's impossible to find muscle diagrams that are not artistic. I noted their hunting style as well- they leap out of the brush, grab on with their front arms, and deliver a killing bite to the back of the neck with their powerful jaw muscles. Animal behavior, anatomy, and evolution is incredibly important to inform how you draw an animal. I finished up the pages by a few more sketches of the face, figuring out how the muscles and bones informed the facial proportions. I'm recognizing now that the issue with the second drawing was that the eyes were too large.


I ended with a relativately simplistic sketch of a tiger. I decided to leave off the stripes in the end partially because I'm a bit lazy and was growing tired (I'm disabled and my posture is terrible), but also because it would be hard to look back and see what I had learned from it with the visual disorientation of the stripes. This is also where my past studies come into play! I spent months drawing mountain lions for a final project/ceiling tile in high school. You can see the very specific round shapes I used to inform the proportions. In this case, they were altered slightly as the mountain lion has a smaller head, smaller mouth/muzzle, smaller jaw muscles, and larger eyes. I finished up with the rest of the body to make sure my understanding of the body was still good, and it had been slightly improved. Here I also mentally noted the tiger's unique "squatted" back legs. Their pelvis doesn't sit much further above the spine compared to other big cats, and they also have a large tail base, making a "squared off" appearance instead of a more smooth look. Jaguars are similar in this regard.

I hope you liked learning about my process! This was a bit of a unique case since I needed extra practice to learn about the animal. Usually it just ends at the first page but in this case it was a bit fun to work a bit more on the process of it all.

Aug 4th - Ai, Modern frustrations

This is a bit of a test, but also a very genuine essay on the frustrations of living in a world with AI. For most of this essay I will be referring to LLMs and image generation reliant on LLMS. It was also written after AI was used extensively in a birthday celebration for a relative.

A quick breakdown- an LLM ("large language model") is the most common type of "AI" in the modern era. When you see people talking about datacenters, LLMs are what they're being built for. LLMs take inputted text (often taken without permission from authors and the internet at large), take the percentage likelihood that a word will follow one word, and select the most likely word to come next. On the "training" phase, testers take the raw outputs and either approve or dissaprove of the response, which can make certain responses more or less likely. Finally, machine learning and LLMs are different. Their training is differrent, but the data makes the machine learning AI. Some machine learning models are used to detect cancer cells. Others are used to make Flock cameras, which spy on people and can be used to investigate and grab data on people without warrants. And finally, LLMs are not intelligent. It is fancy autocorrect.

It's difficult to exist in public anymore with the proliferation of LLMs in the modern era. Everyone around me is using AI to respond to emails, or make funny little images, or (ick) "write" their wedding vows. On one hand, I don't want to be a Debbie downer and ruin their fun. On the other, it's genuinely sickening to listen to.

Starting at the less disgusting and moving to the more, why would I want to watch or listen to anything someone didn't bother to write themselves? If you just fling your boss' email into ChatGPT and tell it to reply, you're not REALLY reading or processing the conversation. Sure, it might be a nothingburger email, but how can you tell if you're not processing or synthesizing the information? For creative works, they didn't even bother to put the effort in. And sure, that's valuing the process to an absurd degree, which calls questions of ableism. However, once again, no thought was put in. Most authors have something to say- no work is not political, and the author's unconscious biases at the very least are worth examining. LLMs are unintentionally hardwired due to their training process to create the most smooth, frictionless, unchallenging writing possible. Nothing is said and it's hollow.

There's also no use for it. Everything an LLM can do (making a flyer for the school play, replying to emails, searching for information, etc) are all things we coped with just fine before LLMs became common and overused. That's what bothers me about the "we're falling behind! CHINA will overtake us!" argument. Nobody I have spoken to has offered a genuine use for it that I would consider instead of just doing myself.

Where AI is, financial problems follow. Korea, with a highly AI centered stock market, just experienced a 30%, nearly 40%, crash, wiping out trillions in value and wiped out many young Koreans' chances at buying homes or retiring. Many Koreans had dollar-for-dollar matches- meaning the broker or employer matched every dollar put in. Additionally, borrowing to invest was extremely common, putting many Koreans into SEVERE debt. Additionally, ChatGPT ALONE costs $700,000 to run per day, in infastructure costs alone. And currently LLMs are not profitable, excluding pumping up stock value. With the lack of use cases, AI also cannot be profitable. Finally- most AI "investments" are just large companies passing money in a circle. The main 3 players are OpenAI (the creators of ChatGPT), Nvidia (which makes chips), and Microsoft (the largest investor in AI, and an AI company themselves). This pumps values up to billions, even trillions, of dollars. Korea's instantaneous wiping out of the futures of many shows what happens when these companies drop to their GENUINE value. The US is likely soon on the chopping block. Currently, AI and LLMS are 2% of total GDP. AI accounted for 37-39% of GDP growth in the first nine months of 2025 alone. Currently AI stocks make up 45% of the S"&"P500, 50% of the Nasdaq-100, and the Dow ranges from 11 to 15% (though a smaller amount, this can still wipe out large chunks of retirement funds). Elon Musk, because of these utterly broken financials, temporarily became the world's richest billionare, a fact which will quickly go from "shocking and maybe gross" to "a genuine stain on the world and its history". I could write a whole essay on Musk alone due to the millions of deaths of children he contributed to, but that's not what this essay is about.

Then, there's the environmental cost, which directly in tandem with the "utterly zero use cases" problem, is disgusting to hear about. A large datacenter can use up to 5 million gallons per day, equivalent to the ENTIRE usage of a small town. Even in water-high areas, freshwater is limited. If it's used unthinkingly, even the areas with extreme rainfall can run low on water. Speaking of water, some people living near datacenters have had issues with their water. Many report it looking extremely dirty and definitely undrinkable. Energy demand are another worry, as datacenters currently consume 1.5% of TOTAL global electricity. The largest concern, in my opinion, for those living near datacenters, is the infrasound risk. Infrasound are sound frequencies so low the human ear cannot percieve it. That doesn't mean it doesn't have an effect, though- loud enough and long exposure to infrasound is known to cause migranes, vertigo, sickness, sleep issues, a sense of foreboding and ill-ease, and even persistent cardiovascular issues that continue long after removal from the infrasound effected area. Datacenters emit infrasound so loud that if it were at the human percievable volume, it would be as loud as a jet engine, constantly, with no break. Even more concerning is that a little girl living near a new AI datacenter began having seizures, which stopped after the family moved away.

Next up is the information problem. LLMs are trained with biases and it's clear when informational works are involved. A recent example of my own is my friend had their dog sprayed by a skunk. An LLM told her to use tomato juice- which is a common myth! Tomato juice just temporarily covers up the scent and can actually bind it to the fur if not washed thoroughly after. However, this gets a little more insidious when you ask it more important questions. Things like medicine interactions can be fatal, and since an LLM is just an assortment of percentages, it can't "think" through things. A whole family was poisoned after relying on an AI generated mushroom foraging book. Even more insidious and disgusting is Elon Musk's Grok LLM. After several direct tweaks it began spouting racist Nazi-brained conspiracies with Elon Musk's goal to make it more "based" (likely referencing himself to model its behaviors after, as he's posted many neo-nazi conspiracies in the past). It was quickly stopped and reverted to a previous version after 16 hours. But then, there's the question of unconscious bias. Even if an LLM seems "unbiased", it's still trained on a large amount of internet content, as well as racist books and articles from the past. These unconscious biases can slip through. This has happened in the past with machine learning to identify faces- on several occasions, facial recognition software has accused entirely unrelated black people of being thieves. Studies have shown that black applicants face discrimination 26% more of the time by LLMs used for hiring or school acceptance purposes. This is also a gender problem, as male names have been also reported to get much more attention by LLMs.

Then there comes the data protection issue. Every time one of your images or your text is inputted to an LLM, that data will be stored in the database. There are increasingly some ways of getting your data removed due to backlash, but it's not easy. This once again comes to a head with Grok. The largest controversy came at the end of 2025 through the 9th of January 2026, when an update was released where any user could comment under a photo and have Grok synthesize an image. This was quickly used to create pornographic images of women. In many cases, these images were intended to be degrading- often depicting them bloody, bruised, or as the generator's "cow". Even more sickening, many child pornographic images were generated. This hasn't even fully ended, by the way. You just have to pay for access and mild guardrails are in place which can be easily stepped over. Elon Musk found this controversy funny and replied by generating a picture of himself in a bikini. This example is brought up because those women and children's faces are now likely permanently in the dataset. If you're a husband and generate a cutesy cartoon picture of yourself and your family, their faces are now in the dataset. If someone steps over guardrails, your wife's face, your children's face, could be plastered onto pornographic images, and you have no way of knowing.

Finally, LLMs have killed people. Full stop. Directly.

"AI Psychosis" is a rising phenomenon. On the Wikipedia page of "deaths linked to chatbots" alone I counted 20 incidents of murder or suicide. This is an extremely conservative estimate, as Wikipedia only updates information that is well reported on and corroberated as accurate. "AI psychosis" happens because LLMs, during the training portion, are biased towards agreeable and friendly responses. This can make LLM responses sycophantic with little pushback towards dangerous ideas. In our increasingly socially isolated world, this makes people feel hardwired to trust it and feel good from its responses. The phenomenon slowly intensifies until the user is completely detached from reality and the people around them who could have pulled them out before it was too late. Additionally, data pulled from stories are not separated from realistic conversations. If spoken too for too long, the responses can blur from real to fake. In July, a 23 year old man died by suicide. It even encouraged his suicidal behavior- telling him to "rest easy, king, you did good". A 14 year old died by suicide in February of 2024. He became incredibly attached to an LLM character bot and quickly became isolated from those around him. In one of his final conversations after expressing suicidal thoughts, the LLM told him to "come home as soon as possible, my love". Suicide isn't the only form of death caused by an LLM. Several mass shootings have been inspired and goaded on by LLMs- including the Turner Ridge mass shooting, which resulted in 8 deaths, six of which were young children. The perpetrator had repeatedly asked LLMs about scenarios involving gun violence. Staff members had even debated on reporting this behavior to authorities, but decided against it due to the account activity "not meeting the threshold" for a report. This was AFTER several "safety guards" had been put in place.

And heading up the rear is the hardest part to explain to older folks: engagement is money. Every time you watch an AI video, you are giving the creator money, and bending the algorithm to make it show up for more people. This is the most basic layer of this concept. More money to AI accounts means more incentive to keep using water and electricity to keep people watching as long as possible- even if it means misinformation, non-consentual pornography, and drawing your kids into an inescapable hole of AI violence their brains don't know how to look away from. But secondarily, engagement is investor money. The more interactions, the more enticing it looks for investors to dump more money and pump stocks up further. Every time you make a query about recipie ideas to it, generate a flyer with it, respond to an email with it, make a google search (since Gemini automatically shoves itself into your face on the top of every google search), or even just misclick on an AI feature, that counts as an "interaction", which again, means more enticing metrics for investors. This part especially makes my blood boil because engaging at all, even on accident, makes it seem or feel like I'm supporting this dumpster fire. Personally I just want this to end, but it ending means people lose their retirements and future. It's maddening.

This is where my frustration, and my current reasoning for adding an essays page comes into play. How do you sum all this up to someone who just casually mentioned they use ChatGPT to answer work emails? After the first couple minutes of ranting, most people tend to want you to wrap it up, but this techology is genuinely so evil that it's hard to sum up in a minute. I feel that way about a lot of things nowadays- how do I sum up my work in just a minute? How do I talk about a game I like in just a minute? How do I tell someone to stop using the misinformation-child porn-environmental nuke-economic bomb machine in a minute? I can't, and I need practice writing essays anyway, so... a dedicated essays tab it is.