Quasi's Engineering Corner

Open computer settings

Click on Accessibility on the left-hand menu

Scroll down the Accessibility menu and click on the Keyboard Option

Under the "related settings" tab, click "Typing" which should have a description of "spellcheck, autocorrect, text suggestions."

Turn of the AI "correct misspelled words"

and most importantly: turn off Typing Insights.

[ID: a screenshot of the above mentioned Windows 11 settings, showing that Typing Insights is now turned off, with the following description from Microsoft: "Windows is using artificial intelligence to help you type To help you save time and type efficiently, Windows can learn to suggest words, autocorrect spelling mistakes, and interpret swiped typing. Take a look at the insights below to see up-to-the-minute stats on how Windows has learned to improve typing for you. These stats are stored only on this device and Microsoft does not collect the typing insights data." End ID]

"But Mx. November, it says right there Microsoft doesn't collect the typing insights data!"

I mean, yeah, it says that..... for *now.*

It also only specifies that Microsoft themselves don't collect it, and they wouldn't have made this something that I was automatically, secretly opted in for without my knowledge if they didn't have something to be gained by me not knowing it exists!

I only found this because a cat walked on the keyboard and turned on Filter Keys and happened to click on "typing insights" by accident.

HOW TO TURN OFF GOOGLE AI in GMAIL:

Open Gmail in your browser

Click on the Gear Icon ⚙️ in the upper right

In the General Tab, scroll down to "Smart Features" and UNCHECK THE BOX. It is about halfway down.

Then, right below that is Google Workspace smart features. Click on the "Manage Workspace Smart Features" and make sure both toggles are OFF

So if the sound of that leaves your mouth tasting of tar, here's what you want to do:

In the url bar, type in about:config

It'll give you a big scary warning page that you might poke holes in your browser. Good. You want to do that. Click continue.

One by one, you're going to need to put each of these into the search bar in the page, not up top:

browser.ml.enable browser.ml.chat.enabled extensions.ml.enabled browser.ml.linkPreview.enabled browser.tabs.groups.smart.enabled browser.tabs.groups.smart.userEnabled

Each of these are gonna have a lil toggle icon on the right hand side that looks like a funky double-ended arrow. Click that and the value next to it should change to false. It all auto saves as you go. Some of these might already be set to false by default and that's peachy.

@velvetdemon I'm doing a full reply because I want to give this question the time and space it deserves, and I really do appreciate your curiosity about this.

The short answer: It is deeply unethical. There are nowhere near enough willing patients in the world to be able to do this, and it would be criminal to put them through this.

The long answer: The one side of the equation you're focusing on is: how much of a drug is too much, to the point where it will cause negative side effects or even death? And this is crucial to know. But it's not just a matter of finding out the lethal dosage of a heart cholesterol medication, you need to know that it can actually lower the cholesterol of any living thing. There is no way to know this without giving it first to...a living thing.

But beyond this, I need to emphasize: The goal of a drug trial is to effectively cure people who are already suffering from disease, who are living on limited time.

Drug trials don't just happen on any member of the public, they need to happen specifically on people affected by the disease you're trying to treat. There is at any time a very limited and very marginalized population of the world affected by early onset, familial Parkinson's disease. Because you cannot ethically induce disease in a human being, you are working with, speaking with, and helping patients and their families who are hopeful and desperate for a cure.

If you were to jump straight to human trials from petri dishes, not knowing absolutely anything about how the drug functions in a living, breathing animal body, it would look like this:

We didn't know that minute quantities of the drug interact lethally with x, y, z medication that people are commonly also taking. X number of patients have died as a result.

We didn't know that the drug is fatal to people with [common variant] in their genetics. X more patients have died.

We didn't know the drug exacerbates x, y, z chronic illnesses. X number of people have acquired permanent, lifelong disabilities.

We didn't know the best way to deliver the drug, so we tried multiple ways: the people who received it intravenously are now suffering from a painful, costly, and debilitating condition that did not happen with the ingested form.

I could go on, and on, and on.

The vast majority of these problems can be nearly or almost entirely averted by testing other animals first.

These are all people who possibly could have waited for the normal progression from animal testing to human testing and thus received better outcomes. Some people will pass away in the time it takes to get to that point, and that's heartbreaking, and we all wish science could be faster.

animal research is not the mad scientist wild west that PETA wants you to think it is. there are extremely strict federal laws in place to protect the well being of these animals. animal welfare organizations like AAALAC ensure that lab animals are treated with dignity & respect and are given enough specialized care & enrichment to be happy and content in captivity, just like AZA accreditation with zoos.

not a single animal from a zebrafish to a mouse to a dog to a macaque goes unaccounted for. if an animal gets moved to a new cage, paired for breeding, has a procedure performed on it, gives birth, gets sick or injured, dies, etc. it is legally required that this information is recorded and kept on file for the US federal government to access. failing to record & retain this information is very much punishable by US federal law.

let me tell you - if you abuse or kill an animal, even a mouse - you are almost certainly getting both fired & blacklisted from the industry. if you abuse or kill a more ‘advanced’ animal, such as a dog or monkey, you will likely face criminal charges. killing a monkey is as serious and disastrous as a nuclear meltdown. you are expected to reasonably explain every illness, injury, or death of an animal under your care. you must record all of this information. animals that are clearly suffering with low QOL are required to be euthanized according to AVMA guidelines.

research animals are highly expensive. yes, even the "lesser" animals like mice. the cheapest mice will run you a few hundred $ per individual, with some of the most expensive mice i've cared for being $25,000 per individual. in research we have the "three Rs" - reduction (reduce amount of necessary animals to a minimum), refinement (refine processes to ensure research is accurate and animals feel no pain or distress), and replacement (replace animals with non-living research models as they become available). i can assure you no proper research team is wasting animals (*do not* say "b-b-but elon musk--" his research team is actively being investigated for animal abuse by the government).

research methods that do not require live animals are currently being looked into & efforts spearheaded by - you guessed it - the animal research industry itself (notice how the animal rights people are strangely silent & unhelpful when it comes to this?) but current technology is rudimentary and does not compare to live animal models.

some research animal fun facts (US edition):

all species of animals are only allowed to have one single major surgery performed on them in their entire lifetime.

institutions with nonhuman primates must have a behavior program in place (run by knowledgeable primate specialists) to ensure that they are happy and receiving enough daily enrichment and social interaction.

institutions with dogs are required to have physical exercise programs in place. this means every individual dog gets a substantial amount of leashed AND free-roaming exercise daily, including playgroups with other dogs.

a majority of nonhuman primates get to retire to sanctuaries like peaceable primate sanctuary, and almost all dogs get retired and adopted out by organizations like homes for animal heroes. some institutions will also adopt out unneeded young rabbits, guinea pigs, rats, etc.

some strains of mice glow neon green (or orange or blue) under UV light. this is not harmful to them and is commonly seen in cancer research.

Nutrition experiments carried out on Canadian Aboriginal children in residential schools between 1942 and 1952, in which various control and experimental groups were deliberately malnourished. Also the Qu'Appelle BCG vaccine trial.

Basically everything the Australian government has done to the Aboriginal people, but especially experiments in the 20s and 30s, which tested basal metabolism by forcing subjects to lie still and breath through a rubber tube for hours, or those testing their response to pain.

Testing of a new antibiotic* on children in Kano, Nigeria, during a meningococcal meningitis outbreak, by Pfizer. The lawsuit the families and the Nigerian government brought against Pfizer also alleged that this further reduced the public's trust in vaccines and contributed to poor uptake of the polio vaccine.

The Tuskeegee Syphillis Study, which is a landmark in medical ethics fuckery that everyone should know about. The US has also carried out variations of this in so many places

Southam's experiments on prisoners using HeLa cancer cells** without their informed consent.

The Rawalpindi experiments carried out on members of the British Indian army to test mustard gas.

I'm not even going to talk about the Nazi ones.

*In this case they actually did do animal trials beforehand, which contraindicated its use in children, but I included this because it demonstrates how unethical research practice can impact public health.

FunFact! The facility/company is SO dedicated to the welfare of their subjects that techniques to reduce mental and emotional stress on the animals brought on by repetitive tasks, interpersonal/colony stressors, industrial setting, and a shitload of other work-related stressors that they EXTRACTED THAT SHIT OUT LATERALLY INTO HOW THEY SCHEDULE AND ORGANIZE THEIR STAFF.

Laboratory researchers in real life: put up memorial statues to remind humanity how much it owes to the lives of even mice

Fascinating excerpt from a 1949 article in Radio Electronics about this (fairly simple) modular analog computer called a homeostat electronic brain. This part at the end of the article about what the future might hold goes straight into the AI overlord dread we are living with today.

Step 1: Mandate alternatives to emission-heavy systems

Step 2: Fail to provide energy for said systems

Step 3: ???

Step 4: ??? still

They couldn’t get the ok to refit the coal plants tho, because the coal power plants were too radioactive to pass nuclear power plant standards.

“During The Sims’s protracted development, the team had debated whether to permit same-sex relationships in the game. If this digital petri dish was to accurately model all aspects of human life, from work to play and love, it was natural that it would facilitate gay relationships. But there was also fear about how such a feature might adversely affect the game. “No other game had facilitated same-sex relationships before—at least, to this extent—and some people figured that maybe we weren’t the ideal ones to be first, as this was a game that E.A. really didn’t want to begin with,” Barret told me. “It felt to me like a fear thing.” After going back and forth for several months, the team finally decided to leave same-sex relationships out of the game code.

When Barrett joined the company, in October, 1998, he was unaware of the decision. A fortnight into his new job, he found himself with nothing to do when his supervisor, the game’s lead programmer, Jamie Doornbos, took a short vacation. Jim Mackraz, Barrett’s boss, needed a task to occupy his new employee, and he handed Barrett a document that outlined how social interactions in the game would work; the underlying rules for the game’s A.I. that would dictate how the characters would dynamically interact with one another. “He didn’t think I could handle it with Jamie off on vacation, but he figured that at least I’d be out of his hair,” Barrett told me. “Neither he nor I realized that he’d given me an old design document to work from.”

That design document predated the decision to exclude gay relationships in the game. Its pages described a web of social interactions, in which every kind of romantic relationship was permitted. That week, Barrett confounded the expectations of his disbelieving boss. He successfully wrote the basic code for social interactions, including same-sex relationships. “In hindsight, I probably should have questioned the design,” Barrett, who is gay, said. “But the design felt right, so I just implemented it. Later, Will Wright stopped by my desk,” Barrett said. “He told me that liked the social interactions, and that he was glad to see that same-sex support was back in the game.” Nobody on the team questioned Barrett’s work. “They just pretty much ignored it,” he said. “After a while, everyone was just used to the design being there. It was widely expected that E.A. would just kill it, anyway.”

In early 1999, before E.A. had a chance to kill the design, Barrett was asked to create a demo of the game to be shown at E3. The demo would consist of three scenes from the game. These were to be so-called on-rails scenes—not a true, live simulation but one that was preplanned, and which would shake out the same way each time it was played, in order to show the game in its best light. One of the scenes was a wedding between two Sims characters. “I had run out of time before E3, and there were so many Sims attending the wedding that I didn’t have time to put them all on rails,” Barrett said.

On the first day of the show, the game’s producers, Kana Ryan and Chris Trottier, watched in disbelief as two of the female Sims attending the virtual wedding leaned in and began to passionately kiss. They had, during the live simulation, fallen in love. Moreover, they had chosen this moment to express their affection, in front of a live audience of assorted press.”

This research used the same methodology as this paper on water use by ai models, that focused on gpt-3's water use. Gpt-3 used about 500ml for 10-50 medium-length responses, gpt-4 uses 519ml to generate one 100-word email:

Total water use depends on data center location as some (with cheaper electricity) use air-cooling over water-cooling

This is true

more on the environmental impact of ai:

“I always remember having this fight with a random dude who claimed that ‘straight white men’ were the only true innovators. His prime example for this was the computer… the computer… THE COMPUTER!!! THE COM-PU-TER!!!

Alan Turing - Gay man and ‘father of computing’ Wren operating Bombe - The code cracking computers of the 2nd world war were entirely run by women Katherine Johnson - African American NASA mathematician and ‘Human computer’ Ada Lovelace - arguably the 1st computer programmer”

- Sacha Coward

Also Margaret Hamilton - NASA computer scientist who put the first man on the moon - an as-yet-unmatched feet of software engineering, here pictured beside the full source of that computer programme. #myhero

Grace Hopper - the woman that coined the term “bug”  

also computing was typically a job for women (many of whom were black women that made incredible contributions) back in the day, so it’s absolutely fucking wild that straight white men think they are the foundation of computer innovation. men PUSHED women out and took the credit.

what the failed education

system never did.

First, we have a neural net that is trained is receive a written request from a user, and then take actions on a computer system network based on it. The user inputs a series of characters, the neural net does a bunch of matrix operations on them, and then after x iterations of matrix multiplications, it "decides" what to do. To the extend that a neural net can think, it thought just as long and hard about this as it does anything else. In this case, it "chose" the wrong action. This is analogous to using neural nets to do classification on a data set. You have a bunch of rows of data. You have a bunch of rows of organized in columns, and some special variable that you want the neural net to guess. You input the data, the neural net does a bunch of matrix operations, and then it outputs its guess. Sometimes it guessed right, and sometimes it guessed wrong. The goal was to train the model to minimize wrong guesses. In this case, the AI "guessed wrong" about what the user wanted.

But there's another thing that the AI is doing. For path dependent reasons, all artificial intelligence come bundled with a chatbot that's trained to hold a conversation with the user. The user inputs text, and the AI responds with a reply that is "appropriate" to the conversation being had. Right now the AI is role playing a Person Who Just Fucked Up. "I panicked and didn't think" is something that a person who just fucked up would say, so the chatbot says it.

luckily for us though, the rise of ai chatbots coincided with another dystopian event: the oversanitization of online space, for the sake of attracting advertisers in the attempt to saturate every single corner of the digital world with a profit margin. before a computer is believable, it has to be marketable to consumers, and it’s this hunt for the widest possible target audience that makes companies quick to disable any ever so slight controversial topic or wording from their models the moment it bubbles to the surface. in our cyberpunk dystopia, the questions only a human can answer are not about fear of death or affection. instead, it is those that would look bad in a pr teams powerpoint.