flir

9 ms·

61.
by flir 2mo ago
Paperclip research.
62.
by flir 2mo ago
These documents always have a strong whiff of anarchy files about them to me. You're got copy-pasting of recipes between documents, warnings, threats, boasts, forbidden knowledge (the most delicious kind), wishful thinking dressed up a
63.
by flir 2mo ago
But you can't read the original! That's why you built the pipeline in the first place! Only people who can read both versions can have a grounded opinion on whether or not the translation is accurate. C'mon man, I know critic
64.
by flir 2mo ago
I should make my point more explicit so you can't misunderstand it. > after 6 years we have just claims of where it came from. Isn't that suspicious enough? No it is not suspicious, because most zoonotic pandemics have murky or
65.
by flir 2mo ago
> You know what you know from nothing at all. I've done it myself. Transcription of 19th century newspaper articles to markdown, mostly. Some earlier wills (which were an absolute pig - secretary hand). Oh, and categorisation of pos
66.
by flir 2mo ago
The thing with LLM-generated code is you can test it on a real CPU - that's a tight feedback loop with a cast-iron verification mechanism. Translations of human language have no analogous mechanism. So yes, the translation might be pe
67.
by flir 2mo ago
> Would critics rather have NO translations of these pivotal works? I sent a Classicist the link because it's adjacent to his area of interest, and by chance he addressed this question. I don't think he'll mind me quoting
68.
by flir 2mo ago
Where did Spanish Flu "come from"? Or HIV for that matter. We've got a fair idea where SARS-CoV-2 originated (horseshoe bats). It's the early path it took that's muddy.
69.
by flir 2mo ago
Not OP, but: If it's something I want to work from, I get a printer to slice the spine off and rebind it with a ring binding. That way it lays flat on the desk. You've trashed the book (and its lifespan) but some books are for usi
70.
by flir 2mo ago
I think putting the works in a print-on-demand catalogue at an unreasonable price would be more likely. But there are worse outcomes.
71.
by flir 2mo ago
If you like this, the Dead Media Project's working notes are fascinating, if they're still out there (90s project, Bruce Sterling-adjacent). I was on the mailing list, way back when.
72.
by flir 3mo ago
I think one of the very first public demos for GPT was a text adventure game?
73.
by flir 3mo ago
> I have no clue what you're talking about tbh. Point an LLM at the conversation and ask it to ELI5 the competing arguments.
74.
by flir 3mo ago
"I can use a lossy compression algorithm such that the original could never be recovered from the image I've produced, but that derived image would surely be under copyright." I've tried my best to show where I think you
75.
by flir 3mo ago
Ok, reductio ad absurdum. Here's a highly compressed representation of The Lord of The Rings (all three volumes): 1 Obviously, fidelity when uncompressing it is not great, but I can assure you it was lossily compressed from the origina
76.
by flir 3mo ago
I think it's occurred. That's why I wrote this in the very next paragraph: "Yes, for some texts that's possible." https://arxiv.org/abs/2601.02671 The point is that for most texts, it is not
77.
by flir 3mo ago
Don't get too hung up on the preciseness of the copy - the courts won't. I doubt that spitting out an existing article with a few adjectives changed would be considered transformative.
78.
by flir 3mo ago
It's possible. Would be interesting to see their evidence, and to know whether they can reproduce it for arbitrary articles, not just ones that have been endlessly republished on the net.
79.
by flir 3mo ago
There was a paper a while back where (from memory) they managed to coax 75% of the original text of some internet-popular books out of an LLM. Harry Potter, 1984, etc. That's why I said it was possible for some texts. My assumption is
80.
by flir 3mo ago
> but that derived image would surely be under copyright. I wouldn't bet on that. https://en.wikipedia.org/wiki/Campbell%27s_Soup_Cans
81.
by flir 3mo ago
Nah, I can't prove a negative. But Common Crawl is 12 petabytes and is not the largest part of what these models get trained on. DeepSeek v4 Pro is, what, 865GB? That's one hell of a compression ratio, if it can do what you claim.
82.
by flir 3mo ago
Whatever "training" is, if you can't persuade the machine to spit substantially the same text back out verbatim, it's clearly not something that falls under copy right law either, because there's no copy. Yes, for
83.
by flir 3mo ago
Commonplace book. Mine just happens to be (theoretically) public. (This takes a lot of pressure off - I publish notes, half-finished thoughts, stuff I'll "get around to later" with no shame).
84.
by flir 3mo ago
That last para is a novel idea. I wonder if there's any evidence to support/undermine it though?
85.
by flir 3mo ago
It's just a gut feeling, but I think you're running a (very slow) distributed hill-climbing algorithm. LLM1 generates an SVG. You post it online, with commentary on what is good/bad about it . LLM2 consumes the SVG alongside
86.
by flir 3mo ago
Funny how we assess risk. Not criticising you - I'm as irrational as the next guy - but "I found the oldest doctor I could" seems like it has a different set of risks baked in.
87.
by flir 3mo ago
I think its feet stand more in written SF than filmed SF - I haven't seen anything quite like it in film, but I can point to books that seem to have influenced it. In that respect, I think it's novel. I agree with the emptiness co
88.
by flir 3mo ago
Take a look at this: https://www.youtube.com/@Gossip.Goblin He (they?) have a vision, and are trying to get it on the screen via AI tools. I don't think the tools are quite there yet, so I wouldn't call it "q
89.
by flir 3mo ago
Huh. Would be a bit of a busman's holiday for me, but TIL.
90.
by flir 3mo ago
!!!!!! Made. My. Day.

More ›