It’s not actually clear that it only affects huge companies. Much of open source AI today is done by working with models that have been released for free by large companies, and the concern was that the requirements in the bill would deter them from continuing to do this. Especially the “kill switch” requirement made it seem like the people behind the bill were either oblivious to this state of affairs or intentionally wanting to force companies to stop releasing the model weights and only offer centralized services like what OpenAI is doing.

If you are at the point where you are having to worry about government or corporate entities setting traps at the local library? You… kind of already lost.

What about just a blackmailer assuming anyone booting an OS from a public computer has something to hide? And then they have write access and there’s no defense, and it doesn’t have to be everywhere because people seeking privacy this way will have to be picking new locations each time. An attack like that wouldn’t have to be targeted at a particular person.

Isn’t it risky plugging usb drives into untrusted machines?

I know that’s how it works in the US, but the lawsuit is in Japan, which you always hear about having stricter copyright laws. Not really sure how this one will play out though.

The detail that negative reviews warning of the danger kept getting removed or edited is really sinister

this will force us humans to go actually outside, make friends, form deep social relationship, and build lasting, resilient communities

There is no chance it goes that way, how is talking to people outside even an option for someone used to just being on the internet? Even if the content gets worse, the basic mechanisms to keep people scrolling still function, while the physical and social infrastructure necessary for in person community building is nonexistent.

I doubt the school administrators who would be buying this thing or the people trying to make money off it have really thought that far ahead or care whether or not it does that, but it would definitely be one of its main effects.

Can anyone recommend any cool mods/projects built on top of Minetest?

I bought some cable organizers, power strip wall mounts etc. there. Generally does what it’s meant to and sometimes cheaper than what are clearly the same products on ebay. Did not install an app. I’m willing to overlook some shadiness if the end result is I get things I need for less money.

The output for a given input cannot be independently calculated as far as I know, particularly when random seeds are part of the input.

The system gives a probability distribution for the next word based on the prompt, which will always be the same for a given input. That meets the definition of deterministic. You might choose to add non-deterministic rng to the input or output, but that would be a choice and not something inherent to how LLMs work. Random ‘seeds’ are normally used as part of deterministically repeatable rng. I’m not sure what you mean by “independently” calculated, you can calculate the output if you have the model weights, you likely can’t if you don’t, but that doesn’t affect how deterministic it is.

The so what means trying to prevent certain outputs based on moral judgements isn’t possible. It wouldn’t really be possible if you could get in there with code and change things unless you could write code for morality, but it’s doubly impossible given you can’t.

The impossibility of defining morality in precise terms, or even coming to an agreement on what correct moral judgment even is, obviously doesn’t preclude all potentially useful efforts to apply it. For instance since there is a general consensus that people being electrocuted is bad, electrical cables normally are made with their conductive parts encased in non-conductive material, a practice that is successful in reducing how often people get electrocuted. Why would that sort of thing be uniquely impossible for LLMs? Just because they are logic processing systems that are more grown than engineered? Because they are sort of anthropomorphic but aren’t really people? The reasoning doesn’t follow. What people are complaining about here is that AI companies are not making these efforts a priority, and it’s a valid complaint because it isn’t the case that these systems are going to be the same amount of dangerous no matter how they are made or used.

They are deterministic though, in a literal sense. Rather their behavior is undefined. And yes, a LLM is not a person and it’s not quite accurate to talk about them knowing or understanding things. So what though? Why would that be any sort of evidence that research efforts into AI safety are futile? This is at least as much of an engineering problem as a philosophy problem.

Seems broken

We’re unable to submit your comments to congress because of a problem on our end. We apologize for the inconvenience. Please try again later.

Youtube’s actual website usually doesn’t even work for me anymore. The page takes a really long time to load and often gets stuck and doesn’t load at all.

This often happens to me on Windows with the Index so it might not even be a Linux specific issue

I used to write that kind of stuff for a living when I was really poor and scraping by, it paid by the word and so low that you could realistically only crack minimum wage if you kept typing continuously and didn’t stop to think or do any research.

Microsoft hasn’t detailed ESU pricing for consumers yet, but the company did previously reveal it will offer these extended updates to consumers for the first time ever

They’re actually gonna make us pirate security updates huh

I guess in this case the malware angle means it’s probably better to require signing, since maybe Russia could successfully distribute malicious fake versions of these extensions otherwise. Still, the centralization here is worrying.

I know it says the extension is not available from the Firefox addon site if using Russian IPs, but I wonder if they have also gone so far as to make the browser itself not be able to install them in other ways. I would guess they have not, since that would mean a complicated setup in terms of the signatures, like they would have to have a separate FF version and set of signatures per country, or use a central server to authenticate things rather than client validation of signatures. In that case it would be easier to find the addon file somewhere other than the store and install it, since using unsigned addons requires installing a whole separate version of Firefox.

Even if that’s how it is this whole thing still illustrates that prohibiting unsigned addons from being installed is user-hostile, because on an ideological level Mozilla probably would use that power to advance state censorship if it came down to it.

Ad-libbed slice of life anime

The person who predicted 70% chance of AI doom is Daniel Kokotajlo, who quit OpenAI because of it not taking this seriously enough. The quote you have there is a statement by OpenAI, not by Kokotajlo, this is all explicit in the article. The idea that this guy is motivated by trying to do marketing for OpenAI is just wrong, the article links to some of his extensive commentary where he is advocating for more government oversight specifically of OpenAI and other big companies instead of the favorable regulations that company is pushing for. The idea that his belief in existential risk is disingenuous also doesn’t make sense, it’s clear that he and other people concerned about this take it very seriously.