river-taxbird

[river-taxbird]

Criterion Collection has announced they will be launching a channel on TikTok to reach a new audience with their preservation of classic films. Launching today, Criterion has posted My Dinner with André (1981) as it’s first upload.

[river-taxbird]

Constructionism is the idea that things are social constructs, (Yes, even the thing you're currently thinking of when you read that sentence.) Constructionism tends to piss a lot of people off on the internet, because they hear social construct and think "Not Real."

But of course they are real, we each construct our own understanding of the world through our education and life experiences, and narratives, and that determines the things we think are real and important. While no one philosophy is the right tool for every situation, constructionism is useful for examining the underlying assumptions and ideologies, and narratives around a topic, that people doing more empirical and positivist research on the topic might not consider.

I think it pisses people off because it is a good tool for challenging worldviews, and it is useful for considering and critiquing why people believe the things they believe about the world. Constructionism serves to critique the assumption that research, education, communication, or any action done by humans can be value neutral and free of agendas. Social constructs are absolutely real because of the all encompassing effect they have on people's lives, but if you are not aware of them it makes it harder to consider alternatives.

[river-taxbird]

Bought some new conversion technology today, (the ones on the right.) Now I can convert in any direction. Such things are needed in the transition to the Type C future.

[river-taxbird]

SCAM ALERT: Cryptocurrency related but can affect trans people buying HRT. I was probably 10 seconds away from having money stolen and I know what I'm doing. (Crypto Clipper)

Today I was helping a trans friend order some DIY HRT. A lot of DIY HRT places only accept cryptocurrency for security reasons. I am not looking to promote cryptocurrency or anything associated with it, but if you may be forced to use it for HRT or other reasons, you need to know this.

In general if you are forced to use crypto, you should use the cheapest coin your supplier will accept. If it is cheap, that means there is not a lot of activity on the chain and energy use will be less. I used one called Zcash as it was the cheapest one the site accepted but that's not really relevant.

I used an old coinbase account I had used for similar situations in the past. I was doing it on her PC. I got the instructions to pay on the HRT site, and I pasted the wallet address into Coinbase and just before I hit send I noticed the wallet address I had pasted didn't match the wallet address I had copied.

I looked it up and found this is from a form of malware called a Crypto Clipper, that detects when you have copied a crypto address, and makes you paste a different one so it can steal your money. I am lucky I noticed. To remedy it, I installed the free trial of Malwarebytes on her PC to remove malware, and completed the crypto transaction on my PC, and confirmed that the wallet address matched what the HRT site had given me.

I managed to avoid falling for it but it's such an easy thing to fall for, especially if you have avoided crypto thus far for extremely understandable reasons. Be careful out there! It could happen to anyone.

[river-taxbird]

1/15/2025

[river-taxbird]

Reminder that the output of ChatGPT and other language models are extremely clockable, due to formatting and word choices it makes that don't align with how people normally write for the internet. (This also applies to how people write for essays, but I am talking about posts here.)

If you want to make it look like something a human wrote, you usually need to edit it so heavily it is basically like you wrote it yourself. So what I am saying is, if you are generating stuff with ChatGPT to post on Tumblr, it is extremely obvious, everyone can see what you are doing, and I will IMMEDIATELY block you if I see it on my dashboard. I have been seeing it more often lately.

[river-taxbird]

I was hoping this study would be published in a peer reviewed journal so I could maybe use it but sadly it was done by a job listings website.

It's still probably informative, but you need to make an account on Upwork to even see it which I don't think is going to happen and makes it much harder to use in academia.

When I became freelance, one of my first marketing contracts was fixing my boss' blog posts and articles that he had 'written' with ChatGPT.

It was the single most soul-sucking task I have ever done in my life. I could have ghostwritten it for them faster than it took me to edit it.

ChatGPT would often hallucinate features of the product, and often required more fact-checking than the article was worth.

It is absolutely no surprise that 77% of employees report that AI has increased workloads and lowered productivity, while 96% of executives believe it has boosted it.

The reality is that it's only boosted the amount of work employees have to do which leads to increased burnout, stress and job dissatisfaction.

Source.

[river-taxbird]

libgen getting culled again and this time followed by smokescreen reporting on its alleged uses for ‘ai training.’ depressing as hell.

[river-taxbird]

What does AI actually look like?

There has been a lot of talk about the negative externalities of AI, how much power it uses, how much water it uses, but I feel like people often discuss these things like they are abstract concepts, or people discuss AI like it is this intangible thing that exists off in "The cloud" somewhere, but I feel like a lot of people don't know what the infrastructure of AI actually is, and how it uses all that power and water, so I would like to recommend this video from Linus Tech Tips, where he looks at a supercomputer that is used for research in Canada. To be clear I do not have anything against supercomputers in general and they allow important work to be done, but before the AI bubble, you didn't need one, unless you needed it. The recent AI bubble is trying to get this stuff into the hands of way more people than needed them before, which is causing a lot more datacenter build up, which is causing their companies to abandon climate goals. So what does AI actually look like?

First of all, it uses a lot of hardware. It is basically normal computer hardware, there is just a lot of it networked together.

Hundreds of hard drives all spinning constantly

Each one of the blocks in this image is essentially a powerful PC, that you would still be happy to have as your daily driver today even though the video is seven years old. There are 576 of them, and other more powerful compute nodes for bigger datasets.

The GPU section, each one of these drawers contains like four datacenter level graphics cards. People are fitting a lot more of them into servers now than they were then.

Now for the cooling and the water. Each cabinet has a thick door, with a water cooled radiator in it. In summer, they even spray water onto the radiator directly so it can be cooled inside and out.

They are all fed from the pump room, which is the floor above. A bunch of pumps and pipes moving the water around, and it even has cooling towers outside that the water is pumped out into on hot days.

So is this cool? Yes. Is it useful? Also yes. Anyone doing biology, chemistry, physics, simulations, even stuff like social sciences, and even legitimate uses of analytical ai is glad stuff like this exists. It is very useful for analysing huge datasets, but how many people actually do that? Do you? The same kind of stuff is also used for big websites with youtube. But the question is, is it worth building hundreds more datacenters just like this one, so people can automatically generate their emails, have an automatic source of personal attention from a computer, and generate incoherent images for social media clicks? Didn't tech companies have climate targets, once?

[river-taxbird]

I had read some of this before but forgotten about it. This reminding me is great because this might actually be genuinely useful for the PhD proposal I'm writing.

The problem here isn’t that large language models hallucinate, lie, or misrepresent the world in some way. It’s that they are not designed to represent the world at all; instead, they are designed to convey convincing lines of text. So when they are provided with a database of some sort, they use this, in one way or another, to make their responses more convincing. But they are not in any real way attempting to convey or transmit the information in the database. As Chirag Shah and Emily Bender put it: “Nothing in the design of language models (whose training task is to predict words given context) is actually designed to handle arithmetic, temporal reasoning, etc. To the extent that they sometimes get the right answer to such questions is only because they happened to synthesize relevant strings out of what was in their training data. No reasoning is involved […] Similarly, language models are prone to making stuff up […] because they are not designed to express some underlying set of information in natural language; they are only manipulating the form of language” (Shah & Bender, 2022). These models aren’t designed to transmit information, so we shouldn’t be too surprised when their assertions turn out to be false.

ChatGPT is bullshit

[river-taxbird]

My mum is a teacher for teenagers in a rural community in Ireland. A lot of her students are from farming families and the ones that are apparently play farming simulator all the time. I definitely think there's a market for kids in farming families too.

just learned about farming simulator

I mean, I already knew about it, but I just learned about it

Did you know that the target audience for Farming Simulator is actual real-world farmers? Because I didn’t. I just assumed that farmers probably don’t want to go home from a day of farming to do some (presumably highly inaccurate) virtual farming?

Like, imagine if the target audience for Power Washing Simulator was actual professional power washers.

Farming Sim gets sponsored by companies and shit to put ads in their games. But since the game is for farmers, all of the ads target farmers. Advertising products that, realistically, only farmers would be interested in. Aka John Deere tractors and shit.

There’s a fucking farming sim esports league. Where do they play? Agriculture conventions. not gaming conventions. agriculture conventions.

[river-taxbird]

Remember that Machine Learning, (One of the things colloquially referred to as AI) actually IS a legitimate and useful academic field within computer science. OpenAI's products and their many clones are a bastardization of it.

[river-taxbird]

[river-taxbird]

Remember that Machine Learning, (One of the things colloquially referred to as AI) actually IS a legitimate and useful academic field within computer science. OpenAI's products and their many clones are a bastardization of it.

[river-taxbird]

Nintendo Power issue 113 (October 1998)

[river-taxbird]

Constructionism is the idea that things are social constructs, (Yes, even the thing you're currently thinking of when you read that sentence.) Constructionism tends to piss a lot of people off on the internet, because they hear social construct and think "Not Real."

But of course they are real, we each construct our own understanding of the world through our education and life experiences, and narratives, and that determines the things we think are real and important. While no one philosophy is the right tool for every situation, constructionism is useful for examining the underlying assumptions and ideologies, and narratives around a topic, that people doing more empirical and positivist research on the topic might not consider.

I think it pisses people off because it is a good tool for challenging worldviews, and it is useful for considering and critiquing why people believe the things they believe about the world. Constructionism serves to critique the assumption that research, education, communication, or any action done by humans can be value neutral and free of agendas. Social constructs are absolutely real because of the all encompassing effect they have on people's lives, but if you are not aware of them it makes it harder to consider alternatives.

[river-taxbird]

really disheartening to see how much eco-fascist and eugenicist bullshit has embedded itself into writings about human relationship with nature. I was looking at a copy of a book in the library a while back called Humans Vs. Nature and found this (Discussing early human migrations in the Paleolithic)

To my great dismay, I did not record the source for this claim, But I found these pictures again, and of course I think...How do we know that?

How could we know that Paleolithic hunter-gatherers deliberately controlled their populations by periods of abstaining from sex? That would be incredibly hard to support using archaeological evidence. It seems easier to support infanticide using the archaeological record, so I was not initially troubled by that.

The author is also stating that Paleolithic humans killed their disabled. I have been searching high and low for evidence to support this claim and the closest I've come to any evidence regarding disability in the Paleolithic is this book chapter discussing whether or not it makes sense to assume compassion existed in pre-history. This book chapter gives the impression that the research has been...really dismal.

The two sides of the debate are essentially, "humans probably cared for their disabled in prehistory, because pathologies and injuries are common and they would have needed some kind of care" and "well maybe those people could survive just fine on their own and that's why they lived. We can't prove they were actually disabled."

Not an anthropologist, but I think it's pretty stupid to position a compassionless society as the "null hypothesis," especially based upon chimpanzees. Why would Paleolithic humans be more behaviorally similar to a relative separated by 5 to 13 million years of evolutionary divergence, than to their own descendants a mere few thousand years later????????

But the claim in Humans Vs. Nature isn't just that disabled people weren't cared for, it's that they were deliberately "eliminated," which is a statement with a much higher burden of proof. You would have to find the remains of disabled humans from that time period with clear evidence that they were killed because they were disabled, and you would have to observe this consistently in many sites, to come to the conclusion that it was a cultural norm.

We have many examples of elaborate, seemingly honorable burials for people that were apparently disabled and would have lived a long time with their disabilities. Nothing I've read has mentioned an archaeological record of killing people for being disabled, which would be a glaring oversight, unless it didn't exist, which I'm pretty sure it doesn't.

How did we get to the point where this kind of fucking bullshit sounds so plausible and correct that it makes it into a best selling book without anyone looking it up to see if it's true.