You’ve got to be careful with rolling your eyes, because the parallelism of the two eyes means that the eye roll can be twice as powerful ^1

(1) If measured against the silly baseline of a single eyeroll

I read this article a few months ago that I found quite interesting: Original link Unpaywalled link

The manufacturer in this blames the DEA, and whilst I don’t trust a pharmaceutical company to do anything other than ruthlessly maximise profits, in this case I’m inclined to believe this depiction of the DEA as the overly persnickety bad guy (because I have even less reason to trust the DEA)

You might enjoy Freya Holmér’s videos - I mostly know her for her excellent mathsy video essays, but she has loads of videos about “maths for game Devs” that might be useful.

When I was at university, the student union had a small fund for creative projects that weren’t related to your degree. Many of the people who applied for cameras also included Adobe licenses on their funding application, because many of them were new to film or photography so they defaulted to what is “industry standard”, because that’s what the majority of online tutorials are available for.

Fuckin’ savage.

That’s a hilarious reply, and I can’t think of a witty response to match it

Yeah, I’m super salty about the hype because if I had to pick one side or the other, I’d be on team “AI is worthless”, but that’s just because I’d rather try convincing a bunch of skeptics that when used wisely, AI/ML can be super useful, than to try talk some sense into the AI fanatics. It’s a shame though, because I feel like the longer the bubble takes to pop, the more harm actual AI research will receive

Eh, it depends on what we count as “AI”. I’m in a field where machine learning has been a thing for years, and there’s been a huge amount of progress in the last couple of years[1]. However, it’s exhausting that so much is being rebranded as “AI”, because the people holding the purse strings aren’t necessarily the same scientists who are sick of the hype.

[1] I didn’t get into the more computational side of things until 2021 or so, but if I had to point to a catalyst for this progress, I’d say that the transformer mechanism outlined in the 2017 paper “Attention is all you need”, by Google scientists.

I mean, racism has as much reason to exist now as it ever did. “I’ll protect me and what’s mine” has been the dividing line between species for thousands of years, and we have to choose whether we’ll continue it. A “Kill or be killed” mindset might keep you safe, but you’ll never know if the person you killed did indeed mean you harm, or if you could’ve instead lived without killing, and broke bread with a rival. The logic still applies

Yup. We’re least safe when we most think we’re safe. Architects of self fulfilling prophecies

I always experience a weird sense of sonder^[1] when I see spam bots, because even if it’s a huge, highly automated network of bots, there’s at least one human involved if you could trace the bots back to their source.

I wonder what kind of person they are, and how the hell they make a profit from the spam. They surely must have a non-zero return rate, but how well does it work? Are they making bank from the task, or barely scraping by? Are they surviving on the scam income, or is it just a bit extra on top of a regular job? Do they enjoy the work, or are they as much of a wage-slave as I am, just in a different way. Were they themselves a victim of some scam, perhaps? I’ll never know, hence the sonder.

[1]: Sonder, noun, neologism. “The feeling one has on realizing that every other individual one sees has a life as full and real as one’s own, in which they are the central character and others, including oneself, have secondary or insignificant roles” (dictionary.com)

The words it chooses to capitalise is funny. For all the training on online stuff, it doesn’t get that it should be calling you a fucking IDIOT for even asking such a BASIC question about grilled cheese!

I personally don’t have nearly as much of a problem with that than I do with Reddit making AI deals. I’m still not keen on the idea of having anything I interact with being scraped for training AI, but aside from only interacting in closed wall spaces that I or someone I trust controls, I can’t change that. That’a not great for actually interacting with the world though, so it seems that I need to accept that scraping is going to happen. Given that, I’d definitely rather be on Lemmy than Reddit.

And this way, who knows, maybe we’re on our way to the almost utopian “open digital commons”

I was talking to a friend recently about this. They studied medieval English and aren’t especially techy, besides being a Millennial with techy friends; I said that merely knowing and using the term LLM correctly puts their AI knowledge above the vast majority of people (including a decent chunk of people trying to make a quick buck off of AI hype)

Oh no, that’s horrifying.

Man, that’s a really dumb story that I find really relatable despite not having had any experiences like that. It feels like it’d be very in character for me though. Thanks for sharing, it helps me feel less silly in the various times where I’ve messed up (of which I am struggling to recall specific examples, but whatever brain part is responsible for embarrassment can remember, apparently)

I learned about this a while ago and meant to check it out, but forgot to. Thanks for the reminder!

I agree. I’ve been trying to move towards open source software for ideological reasons, and I’m astounded by how much my quality of life has been improved. I end up sounding like an evangelist for software that I love because I’m so shocked it’s not more well known.

“analyse your scan for content to suggest a file name”

That’s a ‘Yikes’ from me

That reminds me of a fairly recent article about research around visualisation systems to aid with interpretable or explainable AI systems (XAI). The idea was that if we can make AI systems that explain their reasonings, then they can be a useful tool, especially in the hands of domain experts.

Turns out that actually, the fancy visualisations that made it easier to understand how the model had come to a conclusion actually made subject matter experts less accurate in catching errors. This surprised researchers and when they later tried to make sense of it, they realised that they had inadvertently dialled up people’s likelihood to trust the model because it looked legit.

One of my favourite aphorisms is “all models are wrong, some are useful.” Seems that the tricky part is figuring out how wrong and how useful.