AI as it currently stands works really well in conjunction with a knowledgeable human; expecting it to work perfectly without human guidance is unreasonable. In this regard, much of the marketing hype is the real failure -- and that is on the humans who are generating the hype without conditioning the expectations.

Don't blame the tech for the failure of the developers and operators: Acknowledge the developers and operators as the weak point of a still-developing technology.

Microsoft forcing the Year of the Linux Desktop; who ever would have guessed?

Hmm. I have to admit, the only time that happens to me is when I screw up and there are long running tasks in my process queue.

I'm a SN admin, and I really like SN, so I gotta admit my bias here but... I think it's likely that whoever designs and manages your SN instance doesn't know what they're doing... or didn't, when the system was set up 5-10 years ago, and you're living with old technical debt. I'm not even that great of an admin, and the longest delays we have in our instance are a few seconds... but usually under 1 second for an action like editing or creating tickets.

But it's not difficult (because SN makes all the code and triggers to do heavy customization available) to screw over the performance by making poor choices, and having a ton of synchronous code run every time an action is taken. That's both the glory and the curse of ServiceNow... before it was an ITSM application, it was a rapid prototyping platform. So all that power to code it to do whatever you want is lurking, just beneath the surface, ready to tempt a naive system architect over to the dark side of adding code that runs to every action.

That's only true when the content isn't created by the platform holder themselves.

If I run a forum, you can't sue me for what the forum posters post; you have to sue them. But if I post something (say, by running an AI that creates content), then I can be sued for what the AI I operate posts. If I'm the platform holder, then I can be sued. Section 230 is not really relevant in this case, because when Google posts their own content the S230 safe harbor provisions just don't apply. They're not rehosting content, they are creating it.

You are not wrong. But the gulf between theory and reality is vast, and multiple ISPs have been proven to profit from anti-competetive QoS mismanagement at the detriment of small services and the consumer.

Netflix Agrees to Pay Comcast for Smoother Streaming
EFF Confirms: T-Mobile's Binge On Optimization is Just Throttling, Applies Indiscriminately to All Video

It's not about bandwidth management. That's fine. It's about discriminatory bandwidth management, where some apps (perhaps the ones that pay a kickback to the ISP) get priority over others, allocated a larger share of that limited bandwidth. ISP side QoS can be a good thing in theory, in practice it's just another avenue for them to squeeze more money from the customers and from the service providers by paying for the privilege of being on the 'fast' QoS lane.

It's not really that large of a leap. The issue is the alignment problem... it's much harder than people expect to have the alignment of an AI match the goals of the humans training / building it.

It's not just difficult... we actually don't know how to accomplish it yet. As in, it's an ongoing, persistent problem that's reared its head in many ways. The 'black founding fathers' in Gemini is a current example of the problem rearing its head. Google said 'add more diversity to photos' and the AI complied... but not the way Google intended. Right now it's a silly mistake in generating images... but it's an example of our inability to actually communicate with AI models in a safe and predictable way.

Today it's black founding fathers... but we're creating militarized autonomous drones, right now. This isn't the future, this is the now, and without fixing the alignment problem it's very easy to jump from 'whoopsie' to 'oh fuck'.

Well, there are a lot of people who *do* value it, even if you do not.

How else should people stay in touch with older family members and IRL, though distant, friends? Telephone? Email, which they never set up on their new phone?

So you're saying that the EFF hasn't been the same since 1990? That's some fucking long-ass complaints you have. I've seen a lot of good happen in the past 30 years, so it sounds a lot like you're snubbing good because it ain't perfect.

You misunderstand the issue.

Facebook isn't targeting people based on who it thinks will click. It's the advertisers who are illegally targeting specific protected classes of users and excluding others using the filtering tools that Facebook provides. The lawsuit is over insurance, but this has also come up with housing and other classes of discriminitory conduct.

The advertisers want to target profitable classes of individuals, and ignore unprofitable classes. Men are cheaper to insure than women, so if they can only advertise to men they can bias their enrollments and boost their profits. This is similar to landlords discriminating based on age or racial stereotypes.

So this is different than other algorithmic bias issues. This is about the explicit tools that facebook gives advertisers allowing them to discriminate, even when the thing being advertised has legal restrictions on discrimination.

Did you miss the part about them having servers all over the world? It's the operators that were in China. And you can get out via VPNs, just not reliably or quickly. Don't need a ton of bandwidth to administer a VPS.

... if Japanese culture was less exclusionary, bordering on racist. They could supplement their population with immigration. It would be moderately expensive (training and housing), but a lot less damaging to the economy than a population implosion. However, they're such a proud culture that I don't think they want outsiders anywhere but the tourist traps.

I think this comment is a bit at odds with the usual /. opinion.

WotC literally open sourced the rules of the game and released it under a sharing license for 20 years, and a vibrant ecosystem of both free and not-free content was created around that open, royalty-free ruleset.

Now they're saying "Hey, we're going to retroactively de-license that content you all have been using, and make it closed because you're making too much money and we're not making enough." This would be analagous to Linus Torvalds unlicensing his contributing to the linux kernel because he wanted a share of the money Google made from the linux kernel underlying Android and their enire cloud platform.

WotC has greatly benefitted from the Open Gaming License (OGL) and the thriving ecosystem that has built up around their d20 system. You paint this as a company wanting to benefit from their own work, but that's the stupid part: they are. The OGL is actually one of the most powerful modern corporate success stories for the benefits of copyleft licensing, and now they want to kill the golden goose.

They're free to release new content under any license they want. If they want to be stupid, and go back to a restrictive license for their new content, more (stupid) power to them. But allowing a large ecosystem to build up around their open license would be like a city charging $1 for perpetual land leases, and once a big, thriving, revenue-generating business sector develops changing their mind and saying 'Well, we didn't really mean perpetual leases...'

They should reach out to her for assistance with their physical copy recompression.

Because cleaning efficiency is more important than heat transfer speed for this use case. If you make it hard to remove tea stains, hard water scale, or whatever then you have a worse kettle even if the theoretical efficiency of boiling is higher. Nobody with a hot watter kettle is crying over 13c a year in wasted electricity due to slightly lower heat transfer properties.