Just because an LLM sounds smart and human-like doesn’t mean it will magically solve climate change after being directly implicated in resource consumption we know from actual scientists, today, will make the problem worse.

They don’t believe it.

They just think their investors will.

I’m 100% in favor of requiring models to be open source. That’s been my belief for a while now, because clearly, if someone wants to make an AI model off the backs of other people’s work, they shouldn’t be allowed to restrict or charge access to those models to the same people who had their work used, let alone other people more broadly.

It’s mildly effective in the sense that it will decimate click-through rates, but if enough people did it, they would start filtering by IP, and you’d need to change how many ads it clicks on so it looks more human.

It also still gives advertisers your data, since it still has to load the ads on your system to click them, so it’s not as privacy-preserving as a full-on adblocker that outright blocks every advertisement and tracker related network request in the first place.

This can actually be true, depending on how the system is configured.

For instance, if you and someone else use the same locally-hosted Stable Diffusion UI, both put the exact same prompt, and are using the same seed, # of steps, and dimensions, you’ll get an identical result.

The only reason outputs are different between prompts is because of the noise from the seed, normally randomly set between generations, which can be easily set to the same value as someone else’s generation, and will yield an identical result unless the prompt is changed.

I’m glad about this, honestly.

If you want to use an AI model trained on vast sums of publicly posted work, go for it, but be ready for the result to be made into a truly public work that you don’t own at the end of it all.

Better than completely allowing capital to do whatever it wants without even attempting to push back.

As Cory Doctorow put it, “An app is just a web-page wrapped in enough IP to make it a felony to add an ad-blocker to it.”

The highest usage of ad blockers happens within the age range of 18-24, which categorically includes Gen Z.

The second highest age range is 25-34, and the third highest is 12-17, which is also included in Gen Z.

That said, I would argue that, while knowing how to use a smartphone doesn’t make you tech savvy, knowing how to use an ad blocker doesn’t either. It’s as easy as installing an extension.

I suppose they could, but even cold storage has a cost, and with the scale Discord’s operating at, they definitely have many terabytes of data that comes into the CDN every day, and that cost adds up if you’re storing it permanently.

I also think the vast majority of users would prefer being able to upload much higher resolution images and videos, to being able to see the image they sent with their messages a year ago. I don’t often go back through my messages, but I often find myself compressing or lowering the quality of the things I’m uploading on a regular basis.

They could also do the other common sense thing, which is to, on the client side of things, compress images and videos before sending them.

The thing is, I did have encryption keys set up. The problem was that Element would repeatedly forget the very encryption keys passed by the other user, and would then have to request the keys again. Any historical message history would be permanently encrypted forever, and wouldn’t decrypt with the new view key.

After this happened about 4 times, I stopped using it, because it was impossible to maintain conversations for longer than 1-2 weeks before they’d inevitably be lost, and I’d then have to spend about an hour waiting for Element to receive the new encryption keys from the people I was contacting, even when they were already actively online.

I have no clue what was causing it, but it happened on multiple accounts, on multiple devices, all the time, and there was no conceivable fix. I’m not sure if this is fixed now, but I haven’t had a good reason to go back, especially with other encrypted messaging options out there.

Look, I’m as upset as you are about the enshittification of everything, but this is a bit too far.

There was always legitimate issues with Discord’s storage management, and they at least seem to be taking it seriously now.

I’m not a massive fan of Discord, but this is a bit of an overreaction.

For real.

I emailed them once asking about how they were complying with GDPR regulations if they didn’t allow users a way to delete all their message details, and didn’t even have a procedure for GDPR requests, only their standard, much worse privacy-wise account deletion process. They claimed it was because they had a legitimate interest to keep any messages not individually deleted, so the chats would still look coherent after an account was deleted.

They only delete your message if you delete it individually, so naturally, I was concerned, since you can’t delete messages in a server you were banned from, or left, and Discord provides no way for you to identify old messages in servers you’re not currently in.

They eventually, supposedly, sent my concerns to their data privacy team.

They were then sued for 800,000 euros about a month or two later.

They still don’t allow you to mass delete your message data. They really want to hold onto it for as long as they can.

Matrix is nice, but it’s still very bad UX wise.

I’ve used it on and off for years now, and about 2-4 times a month it loses my chat view encryption keys, and loses me my entire chat history. It also regularly has sync issues between devices signed into the same account, and is relatively slow sometimes to send messages.

Of course, that’s just my anecdotal experience, but I’ve tried many messaging platforms over the years, and while Matrix (and multiple of its clients, primarily Element) is the most feature-complete compared to Discord, it’s nowhere near properly usable long-term for a mass-market audience.

Same here, honestly. I would have thought they’d say something like “hey, we’re going to delete anything 1 year or older starting next month, and reduce that amount slowly down to 6 months with time” just to give people a general warning in case there was anything they were storing through Discord that they wanted to keep.

There’s also just a ton of optimizations they could have done. Are people repeatedly uploading the same file, with the same name and contents? merge them into one CDN link. They’d probably save hundreds of terabytes of data just from reposted memes alone through a hash matching algorithm.

The Open Source Initiative has a giant list of licenses that anyone can use to make their works fully open-source.

Some are just for code, but I’m sure they could be adapted to things like medicine, if needed.

I would be at least a bit worried too, but unfortunately the only reason this exists is because corporations decided to wall off access to producing their drugs legally so they could continue to exploit vulnerable people for profit.

For a lot of the people using this tech, it’s the only way they’ll get life saving medication, and without it, they’ll die. If that’s the kind of gamble they have to make, a possible risk of impurities or negative reactions is better than the considerably less desirable option of death.

Technically, drug dealers are using the tech (more specifically, other people are using it, then selling the product to the drug dealers, who then sell it to their customers as a ‘service’ included with the drugs)

The thing is, they’re not doing it to make stronger drugs, or for PR purposes. They’re actually adding pre-exposure prophylactics (PrEPs) into their heroin, which then creates the side effect of preventing the contraction of HIV from the needles. (referenced about 1/3rd of the way down this article)

If people are already going to be addicted to these drugs, them not getting HIV from it is just one harm reduction measure that can reduce their risk of serious, permanent illness.

Well that’s the coolest part about this, everything is based on the existing research.

The drugs they’re making are the exact same chemical compounds formulated by the drug companies, and contrary to popular belief, the compounds can actually be relatively simple, it’s the process of finding which compound that takes the most money from R&D.

So if you have 2-3 very standard chemicals, with well known reactions and outcomes, and you have the exact blueprint of what the final result should look like, and you can chemically test it afterward to see if it combined as expected, then anyone who has enough reason to use this instead of traditional means (i.e. being priced out of lifesaving medication completely) can be reasonably confident it will work.

And it’s only made more inspiring by the fact that he has his own personal history with the pharmaceutical industry that didn’t work for him.

I found another article on him and the collective, and there’s this honestly saddening quote:

“A toast to the dead, for children with cancer and AIDS,” Laufer said, raising a glass of bourbon and quoting the hip hop artist Felipe Andres Coronel, better known as Immortal Technique. “A cure exists, and you probably could have been saved.”

It’s even posted up on their page for the MicroLab right at the top.