I suspect it’s not dissimilar to the way spam emails are full of typos and grammar errors. You may wonder why they don’t just get those fixed, but they’re specifically to filter out the people who notice them and dismiss the spam, as they (the spammers) are far less likely to successfully scam someone who is offended by the way the spam is written. They are a kind of first level filter.

MS are filtering out the vocal, knowledgable people who will cause problems next time they have some security breach or do something shady around privacy. Convert that relatively small number of people to Linux, and you’re left with a compliant and fully tracked customer base—far more use in the long run.

I guess the company was providing a kind of UBI? Not sure what will happen when all of those non-jobs disappear…

My PhD was in neural networks in the 1990s and I’ve been in development since then.

Remember when digital cameras came out? They were pretty crappy compared to film—if you had a decent film camera and knew what you were doing. I fell like that’s where we’re at with LLMs right now.

Digital cameras are now pretty much on par with film, perhaps better in some circumstances and worse in others.

Shifting gear from writing code to reviewing someone else’s is inefficient. With a good editor setup and plenty of screen real estate, I’m more productive just writing than constantly worrying about what the copilot just inserted. And yes, I’ve tested that.

Surely boilerplate code is copy / paste or macros, then edit the significant bits—a lot less costly than copilot.

Just use Reader view or whatever that’s called in your browser. I use Arctic for Lemmy on iOS and it has a ‘default to reader’ for opening links. Can’t remember the last time I saw a paywall. There’s one news site that doesn’t work but it’s pretty obvious straight away.

Partially. The summary isn’t quite in line with the detail:

Android is the only operating system that fully immunizes VPN apps from the attack because it doesn’t implement option 121. For all other OSes, there are no complete fixes. When apps run on Linux there’s a setting that minimizes the effects, but even then TunnelVision can be used to exploit a side channel that can be used to de-anonymize destination traffic and perform targeted denial-of-service attacks.

4-5 TOTP apps? So far, when, e.g. Microsoft or Google have insisted use of their own Authenticator app is required, it’s worked fine for me using Ente Auth or similar just by entering the code / QR.

Hibernate is S4 which is very low power but not zero. Some devices like LAN, keyboard, and USB can remain powered so your battery will eventually drain.

I think maybe a re-read is in order. They’re claiming the new format outperforms the (presumably) old format by 28%, not that the CTR is 28%.

Plot twist—they work for airbus.

Is that a frickin rotating chair?

It’s in the article:

This swapping is powered by an OpenType feature called “contextual alternates,” which is widely supported by both operating systems and browser engines.

“Starlink was not meant to be involved in wars. It was so people can watch Netflix and chill and get online for school and do good peaceful things, not drone strikes.”

https://m.youtube.com/watch?v=Skl71urqKu0

Not sure if this helps, as I’m not 100% certain how I set this up (it was too long ago), but by giving my rpi a hostname I can simply use ‘pizero.local’ from my desktop machine. I don’t run a DNS server per se but maybe look into that kind of dhcp / dns option?

In fact, see this for some similar hyperbole and sentiment.

Definitely LLMs have been over promised and/or misrepresented in mainstream media, but even in the last few months their utility is increasing. I’m a big advocate of finding ways to use them to enhance people (thinking partner not replacement for thinking). They are most certainly a tool, and you need to know their limitations and how to use them.

From experience working with naive end users, they are anthropomorphising based on how the models have been reported and that’s definitely not helpful.

As the models get more and more capable (and I’m pretty happy to make that prediction), will they reach a point where they are indistinguishable from the output of a real person? That will give us some challenges. But the interesting thing for me is that when that happens, and the AI can write that report you were paying someone to write, what was the point of the report? You could argue they were some kind of terrible UBI and we’ll end up with just the pointless output without the marginal benefit of someone’s livelihood. That needs a bigger rethink.

Better for one obscure use case? Or just ‘better’? That’s the real issue here. OpenAI have an agenda (publicly, a helpful assistant, privately, who knows…). They’re not really interested in a system that can identify prime numbers.

Ok, I’m not going to go point by point, as this is getting too long. All I’d say is remember where the model for ML came from (McCulloch & Pitts), and that this is the worst AI will ever be.

If this is truly a jump across S-curves in utility, it’s bound to be slightly worse than other methods to begin with. Many of the arguments against the current approach sound like the owners of a hot air balloon business arguing with the Wright brothers.

Yes, just remember to wash your hands 93% more.

I have also invented double-sided paper which holds vastly more information.