Agreed - 8b has enough magic to hold a conversation and do small tasks, such as breaking up a large task or picking out key details, which can then be fed into more small models (maybe even more narrowly fine-tuned ones)

180b isn’t enough to replace all the other pieces of a system that you need for autonomous action or memory

I think 8b models are enough to make AGI possible if we stack them just right. They’re enough to fill in most of the gaps to make practical things too, and they’re not that far off for everything else

I love how he literally named things in his field of view for this. That speech certainly included words

Madison, a former employee, basically described a terrible working environment. She was bait and switched on several aspects of the role, and only found out after giving up her US visa and moving to Canada. She was constantly berated, sexually harassed in many ways, made to manage an only-fans despite making her discomfort known, the expectations of releases were unreasonable and her coworkers sabotaged her by knowingly setting her up to release videos with incorrect facts. Taking any of this to HR, or even Linus himself, ended with her being laughed off or told to stop being a tattle tale

All this is alleged and there’s a lot more, but the kicker to me is the fact she cut her leg bad enough to need staples, because going to the ER was the only way she felt she could get badly needed time off.

That’s what seals it for me - either she’s a borderline nightmare, or the working situation is far beyond the pale. Add in the fact the pressure to release, inaccuracies, and public glimpses of Linus’s attitude, and I’m inclined to believe it’s more true than false

Is it though? That amount of money is meaningless to a company, which Linus loves to talk about these days.

The problem is, he shit all over a startup company, failed to return the prototype after multiple requests, then when called out on it offered to pay for it and phrased his response to make it seem like he hadn’t spent months ghosting them until another YouTuber brought attention to the issue.

Shitting on it - fine, he’s extremely harsh every time it’s brought up, but he can have his own opinion. I think it’s a bad take, he doesn’t even entertain the idea that they might lower the price, improve it to work on multiple models, or maybe this fits a high end niche for PC ricing - it sounds impractical now, but maybe a few sales would be enough for them to make a more practical version

But whatever, I can get over that. The fact that he didn’t say “we had some miscommunication in my team, this is our bad, we’re having growing pains and I never would have sold it if I knew they wanted it back. We reached out to them to make them whole, but we’ll do better” is pretty incriminating.

That’s not owning up to their mistakes - either they knowingly ignored requests to give it back, which is fucked up, or someone made a mistake and he made excuses instead of owning up to it, and tried to quietly bury the problem and fire back on the guy who called them out

You have a point, but it’s kind of like people who resist all gun control due to the second amendment despite the shootings going on

Yeah, if life continues on as-is, the argument has little merit. On the other hand, in the case of the second amendment, we have fascists making a credible move for control.

In the case of archiving DRM content, if we have a cataclysm (which seems increasingly likely), then having drm-free, ideally unencrypted, content sitting on random hard drives might end up making an enormous difference in a lot of lives

Or even without a cataclysm, just general enshittification might end up destroying the gaming and media industries - passed around old games might be the seed for the next generation of tech-heads. I started my path by jailbreaking my PSP so I could use custom web browsers and homebrew - spreading these after the Internet is locked down by efforts like kosa and WEI (and whatever comes next) might be the spark that motivates the next generation

To your point, they published a method that could be replicated in less than a week in basically any college or lab in the country.

Fraud makes little sense here - this screams exuberance to me