ollien

I worked in telecom for a couple of years up until recently. There's actually a growing body of self-regulation going on within the SMS industry. Most notably, any business sending text messages has to apply for a "license" to do so, with some pretty strict consent requirements. Violating those requirements comes with heavy penalties, mostly enforced by downstream carriers. If you're curious, 10DLC/A2P are the terms to Google for.

If you ask the FSF, open source is a bigger set than free software, mostly to do with restrictions on the uses of the code

https://www.gnu.org/philosophy/categories.html.en

That's unfortunate. Devices like that are basically impossible to use on certain enterprise networks (e.g. college campuses). There really needs to be an override

I'm no expert, so take what I'm about to say with a grain of salt.

Fundamentally, a LLM is just a fancy autocomplete; there's no source of knowledge it's tapping into, it's just guessing words (though it is quite good at it). Correspondingly, even if it did have a pool of knowledge, even that can't be perfect, because the truth is never quite so black and white in many areas.

In other words, hard.

Does anyone have the recipe on hand? I'm curious what it actually recommended but I couldn't find it with a cursory Google search

I don't have the time to find this right now but there was an example at launch where it very badly summarized pet vacuums or some such, giving blatantly incorrect information about them.

Ditto. I mostly use it when Google (search, not Bard) fails me. I find it's really good at answering questions of the ilk: "I swear there's a function for this in the library I'm using, what's it called again?", or telling me that it doesn't actually exist.

Me personally? Not really. But it's definitely a feature I've heard folks wanting.

Tangential, but my last employer (US based) outsourced L1 IT to a call center in India, and it was maddening. They didn't know very much beyond the script, and often you just had to say the right words to get your issue escalated, but it would always take a day or so to get called back. It drove me nuts as an engineer, but I'm sure it works fine for people who are less familiar with computers.

I've found that the chat agents are much less able to "be a human" and help you out, it feels like talking to a chatbot sometimes. It's a lot easier to get someone to empathize with your problem over the phone, IME

Someone on Mastodon raised a good point that the idea of "changing handles" is incompatible with a lot of fedi, too. I'm curious how they will tackle that problem when they eventually federate

In case anyone, like me, was wondering what "structural regular expressions" are, here's a write-up Rob Pike did in his Bell Labs days

http://doc.cat-v.org/bell_labs/structural_regexps/se.pdf