Found their Github. Looks like it’s a custom Linux kernel…? I dunno, it’s really hard to tell directly; I’m inferring from context clues (like tagging Raspberry Pi, mentioning the open Nvidia kernel modules, etc.).

https://github.com/getumbrel/umbrel

Neat!

So to be clear, these are all papers before the peer review process?

I agree, and I don’t even particularly like the foundational use of AI here. Education should be accessible to all, and it would be cool to see this become a free tool for which OP doesn’t even need AI anymore, because there’s enough contributors and experts.

This software isn’t free. It has a cost. Did you think using the tools provided by billionaires was a gift from them?

Also, it’s hilarious that you think I’m bullying OP. Where in my original comment did I say, “No, don’t use this”? I intentionally skipped over my ethical concerns, because OP is likely going to use it anyway.

They’re providing educational content. You can’t just replace the knowledge of experts creating and checking lesson material with AI and hope for the best; that’s how you end up teaching people wrong information.

Y’all pro-AI goobers are so lost in the sauce, you can’t even take a critique.

$13/mo can be a lot for people, that’s true. There’s certainly a subset where they simply don’t have that extra. I suspect that subset is fewer than people might imagine, but I honestly would rather people have accessible education than not.

For the second part, I think it’s a hard ask for experts to contribute in that way, but hey, people spend their hobby time working on Linux and other OSS, so who knows?

I also have ethical issues with the use of Claude itself (as well as other tools/platforms provided by billionaires), but I think that’s beyond the scope of your original questions.

There are other ML models made from scratch by universities and experts which aren’t beholden to shareholders (and might not be accessible by the public, yet), so if you are determined to continue using AI in this way, keep an eye out for potentially better alternatives.

Get a refund

There’s petabytes of data there dude

The fact that you think that AI has a database of information from which it pulls demonstrates that you have no idea how it works. And you think we’re the blind ones lmao

Ah, right. Because I secretly love it and use it all the time, as evidenced by your very scientific source, “Trust me, bro.” What a tired and boring Reddit cliché.

It’s weird that people are so positive towards AI, even when it’s used to generate something with no understanding of the material. It’s now become a disease to be accepting of anything AI, even if it uses the very thing our oppressors want us to use.

Notice that I didn’t say that $160 wasn’t a lot for people. I said it was a steal for the amount of educational content.

First off: $160 a year for education (about $13 a month) is a steal. I’m not sure why this needs to exist.

Second: ignoring the ethical issues with using Claude so glibly, who is reviewing these lessons for accuracy? Who is making new lessons? Part of Brilliant’s deal is having actual subject experts who are at least reviewing the lessons. No AI model is a replacement for that expertise. You can’t crowdsource that expertise ad hoc, either (the people who use it are the ones learning from it, after all).

Are you just going to hope Claude can clone some facsimile forever?

deleted by creator

I think they might mean that Windows is ultimately a dead end. I agree that its far from dead, but Valve really got the ball rolling with Proton.

Now that there’s a bunch of bigger influencers “trying Linux” as a daily driver and often having a good experience, coupled with Windows/Microsoft shooting themselves in the foot every other month, I think we’re seeing a shift in people’s perception (at least regarding what their possible options are). That said, Linux still has a ways to go until it even breaks 10% desktop share, and I don’t think it will be a threat even to Apple until Linux hits 15%.

Is this some new money laundering thing? Are companies “buying in” to hobnob with other wealthy people? Is this just the latest Country Club or whatever?

I seriously don’t understand why all these companies are throwing money at a project that is almost assuredly going to “mismanage” it.

Yep, I agree with all of that. The original source here is specifically using LoRA, which is why I focused upon it, but Reticulum is certainly more than that.

I think we may be talking about different things. When I originally said “emergency comms,” another way to say that would be “ad hoc communications in a disaster scenario.” I was imagining coordinating teams and disseminating information in a scenario where typical radio stations might be damaged or otherwise unavailable.

It seems that when you say “emergency,” you’re imagining a person who needs immediate medical intervention. For that purpose, I agree that I wouldn’t rely upon LoRA technology unless it was literally the only viable option.

I’m actually interested in your reasoning why you feel that way. Care to expand on that?

Reticulum is a great project, but it’s not a replacement for the internet. LoRA (a potential piece in that chain) has very low data throughput; it’s fine for messages, but you’re not delivering web pages or streaming media with it.

Unless and until the throughput can be increased, Meshtastic/Meshcore is really only a replacement for text messaging, which might be perfectly fine for setting up encrypted comms or rapidly deploying communications in an emergency scenario.

As far as I can tell, Voice doesn’t integrate with anything. You pick a device folder where the audiobook files are, and it’s all local.

Anyone who might know better, feel free to correct me.