My coworker starts almost every Teams message either with “Btw I had Claude do…” or “So Claude and I just…”. If I message him first, there’s a 75% chance the message I get back starts with “Hm, I just asked Claude about this, and…”.

All his PR descriptions, commit messages, and comments are clearly “Claude”.

I’m this close to start reviewing his PRs solely through Claude, and starting the review with “Here’s what Claude came up with in review:”.

The only thing holding me back is that this would mean I’d have to use Claude. So… No.

Also, hard disagree.

“average 50-59 year old weighs 5000 pounds” factoid actually just statistical error. Average person weighs 150 pounds. Pounds Georg, who lives in a cave and weighs over 50000000000000000000000000 pounds, is an outlier adn should not have been counted.

I don’t know. I had never seen another magical girl anime before, and it still stood out to me.

The Dvorak symphony or the Yusuke Kishi novel?

If the former, how dare you.
If the latter, … sorry I haven’t read it. おすすめですか。

Yeah, truly.

It’s a beautiful encapsulation of “what the fuck”.

“Madoka Magica” might as well be my trigger phase by now.

Add “linux” and enjoy the empty feed

Oh, also: you should cross-post this to !sneerclub@awful.systems, they will absolutely love this.

Hah, nice read.

It’s always been wild to me how the idea of “a future AI threatens incredible suffering if you don’t help it come into existence” sparks “oh, then I better help create the torture-AI” in people, rather than “what the fuck, that’s evil and cruel, let’s make sure this doesn’t come to pass”.

But of course, the Rolo folks start from the notion that this AI is 100% inevitable, at which point… There’s no need to sadistically torture people.

In the same vein as “God can’t be all-powerful and all-good”:

The AGI either doesn’t exist, or has no need to torture people.

This started off as a single file in my private nix config, to see if I could get it working at all. In that initial part, some parts were indeed LLM generated (esp. testcases based on my existing intents and failures).

When I noticed that this might actually work and be useful not just for myself though, I moved everything out manually, refactored and cleaned it up, and everything since has just been myself. I guess you’re still right though. I’ll see about adding a disclaimer to the README until I’ve gotten the chance to properly rewrite everything.

Huh, interesting. Does that work with wildcards like “put x on my shopping list”? Also, what are you using for that, if I may ask?

Sorry, I don’t quite follow 😅

What’s the problematic response?

Yay, that’s fantastic to hear!

Also, how’s your experience been with the PE? Getting a readymade device in a nice shell is appealing for sure 😅

Oh, in the demo gif, that’s via a shortcut (holding power for half a second). Sorry, can’t help with wakeword there 😅

Very cool. I’ll definitely look into that, and let you know back here :D

Glad to be of service… 😄

did you consider metaphone matching?

I did not even know about this. Sounds super interesting. Though it seems to be very language specific?

My original intent was to not rely on language specifics. But maybe we could just define additional steps in the pipeline for specific languages. Hm. I’ll have to think about this some more, but it might definitely be a great idea for a future version, so thanks for telling me about it!!

Have fun, hope this works out for you! FYI: you can also use an LLM as an additional fallback (first closest-intent, then on failure, LLM). README mentions it further down on Github.

You mean, the one in the GHSA that was swiftly fixed in accordance with proper protocols?

“Hugely overblown”, you mean.