Alexis Glass

[Alexis Glass @mutagene]

https://t.co/1O5z6228ey 🙏
オートメーション可能な項目、多いんですけど、減らそうとすると、思うように動かなくなります。つまり、モード毎のパラメーターをオートメーション可能にしたいし、自由にモードも変えられるようにしたいわけです。掛け算すると、かなりの数になっちゅうんですが。

[Alexis Glass @mutagene]

@GutenLeinad @jelly_pack I prefer the OP’s dystopia.

[Alexis Glass @mutagene]

@provisionalidea @sean_from_earth This was my thought - needing Greenland to seem worth the cost to MAGA - but they most have known they would lose, and badly. So I guess it's a matter of principle? I guess they're not continental realists.

[Alexis Glass @mutagene]

Bought something for Bandcamp Friday? Get a discount on a quirky resynthesising drum machine: https://t.co/gCYWwhucml (sorry for the double post)

[Ronnie / Rekkerd.org @rekkerd]

Mutagene releases Xoradora resynthesizing drum machine (open beta) https://t.co/c4Jqq068Cv

[Alexis Glass @mutagene]

@halvarflake Reminds me of Spoorloos . I hope he's never wondered if he's a good person or not.

[Alexis Glass @mutagene]

@losslandscape I can't understand Canadians being mad, but I suppose if I had voted for him I might feel differently. At this point I don't think he's even really trolling Canadians so much as American institutions or the seriousness of our timeline. I might agree on the latter point.

[Alexis Glass @mutagene]

@aravosis How are LLMs different from people? I can't see that as a serious question, but trying to: for one thing, you're alive to talk about it - and could even betray your patrons' confidence at some point. Or write a memoir later about the subjective experience of being a ghostwriter.

[Alexis Glass @mutagene]

@izakaminska It sounds like the WSJ is saying that for sufficiently credentialed people, an AI can fully approve a piece for publishing. And some credentialed people apparently don't feel the need to write anything. It is even plausible they may not have read the AI output. Why should we?

[Alexis Glass @mutagene]

@LudvigBerke "Healthcare isn't free, you pay with taxes," they'll say. But refills *are* free apparently - there is no hidden cost for these.

[Alexis Glass @mutagene]

@optionscaller @RnaudBertrand Yes, would agree with the first part - but in fairness, Carney never said it was just one demand or another (even if he did highlight 3). Rather, he said it was the cumulative effect. Maybe some of those demands had been partially attenuated, but together the direction is clear.

[Alexis Glass @mutagene]

@GergelyOrosz I share this fear. It's not enough to just not use AI to write for us. As humans it is natural for us to absorb language and eventually even mannerisms. Bit like that Dr Who episode Midnight. This is a dangerous mind virus; everyone for themselves - but this is a contaminant.

[Alexis Glass @mutagene]

@Roro_X_en @klonopuppy @GregoryConti19 Maybe. “The lyf so short, the craft so long to lerne” - descriptive but also maybe a bit normative?

[Alexis Glass @mutagene]

@Shreyassanthu77 @JLarky This has been bugging me a bit - in Jarred’s post we never saw a like-to-like comparison of bun rewritten in rust for $165k vs the zig build rewritten/optimised for $165k tokens. In that sense, however polite the post was, I feel there was a misleading premise.

[Alexis Glass @mutagene]

@Botbekker1233 @PDXFato As a Canadian I would say we totally are that insecure, and yet... to my continued surprise... apparently many Americans are even more insecure? Weird.

"We're such great hosts, unlike you!" said every great host ever.

[Alexis Glass @mutagene]

"zero glitches" - sounds like a dare?

[Alexis Glass @mutagene]

@Michael_Druggan @JeffLadish Forget ASI for a moment - I'm sure there are humans more disciplined and intelligent than myself could run my life better than I can, but then, where is the point or the dignity?

5 year olds will one day become adults, but we will never be ASI. Why should we want to be NPCs?

[Alexis Glass @mutagene]

@a16z If this were true, wouldn't it already be the case with managers? And it is clearly not. Worst case, take the micromanager - probably more of a liability when dealing with humans, as it is a violation of trust, whereas at worst it is only inefficient when dealing with LLMs.

[Alexis Glass @mutagene]

@grok @CarolingianSon @XFreeze Shouldn't a "truth seeking" LLM, when questioned how another instance came to a given conclusion, preface its answer by explaining that it is neither a single coherent entity nor privy to the exact processes which lead that instance to the conclusion?

[Alexis Glass @mutagene]

@CBeataE In starting to think this is about Gripens and F35s because I just don’t see this double touch as being another 100m anabolic steroids thing.