i really have no patience whatsoever for people who are scared of muslims having elected office
Computational Cosmetologist
@dferrer.bsky.social
ML Scientist (derogatory), Ex-Cosmologist, Post Large Scale Structuralist I worked on the Great Problems in AI: Useless Facts about Dark Energy, the difference between a bed and a sofa, and now facilitating bank-on-bank violence. Frequentists DNI.
for openai to mess up sandboxing this badly indicates openai didn't take the issue seriously, which you expect because they don't take anything seriously for anthropic to mess it up is rank incompetence
I program my puts to execute femtoseconds before The Rip on the UV meta-crystal that houses the exchange. I have calculated the time precisely enough that the trade will almost certainly occur. But my consciousness is limited to meat-speed, and the event is outside my brain's lightcone...
Computational Cosmetologist@dferrer.bsky.social · 23h agoCackling as I exercise my OAI puts femtoseconds before time itself ceases to exist.
Coming soon on my substack: Not a Pop, but a Big Rip
Computational Cosmetologist@dferrer.bsky.social · 23h ago"Implications of the Dark Energy Equation of State for the AI Bubble: The Pop is Coming"
This makes me wonder if there's a better way to conceptualize "annotation" in code or documents than comments, co-located or otherwise. For human use, we've settled on things like "hide the annotation in an IDE unless you hover over the annotated line", but there's no equivalent for agents
em@foremphasis.bsky.social · yesterdayClaude you have got to chill with the comments
I don’t think anyone here is partial to Musk, but “direct to machine code” is going against every discovery in computation including computation itself! For any given problem, there is likely an analog circuit that solves it. Discreteness and “cycles” and “instructions” are abstractions.
conputer dipshit@davidcrespo.bsky.social · yesterdayit’s a genuinely interesting question: is abstraction merely compensating for our limited cognitive capacity or is it in some way part of the logic of information and therefore valuable for any finite cognizer? my hunch is the latter but I admit I haven’t thought about it too much in those terms
"Oho, you believe ChatGPT is competent at something inversely proportionally to how well you understand it." I thought this was a funny line in Jan 2024 too. But now it's pretty much exactly wrong. I trust LLMs the most to do the tasks I am paid for. I am confident my industry is changing.
All time spent is a tradeoff. The time you spend learning something you hate is time taken away from you pursuing something you love. It’s not sad to prioritize: there are infinite things to learn and finite time in which to learn them.
It's worth noting that it's very disappointing to see Hank Green talk that way and folks take it seriously. Because this "dopamine addiction" stuff is trash science. It's Dr. Phil level bullshit. You scrape it like stale gum off the bottom of the same school bus seats you find Gladwell stuck to.
Mike Riverso@mikeriverso.bsky.social · 5d agoThis is exactly how people talk about gambling and drinking.
!["I need to come to terms with the fact that the level of dopamine I've been getting from interacting with LLMs-with doing more and more and more and more-is not healthy for me or good for the world," said Green. "It is careless, and has disconnected me from where people are on this. I was talking about this with [my wife] last night, and she did not pull any punches. Frankly, neither did [my brother]. Neither of them think I've been being healthy. I need to readjust."](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:pytxycyhu6ql3flqs2qmkuub/bafkreicmxurcds4n55kttsu4q7lglli5tmelaerxppn4gc5ruutlkdqrvq)
Tired of RPG companions whose entire personality/ story is a single gimmick. Ok, great, you’re a “Blood Cultist”. But every conversation doesn’t need to revolve around blood and blood cult accessories. Get another hobby.
one of the things i hated the most about "growing up" was learning that companies aren't actually ruthlessly profit-maximizing
Michael Derby@michaelsderby.bsky.social · 6d agoConsider yourself warned: Microsoft is about to supercharge Copilot: “Copilot is evolving rapidly from chat to Cowork to autopilots...one super app spanning both consumer and commercial experience." Via Gizmodo
Other than a handful of exceptions, never assume anything Google makes is a “product” to make “revenue” as opposed to a locus of inscrutable intra-organization rivalry
Brad Ackerman@brada.bsky.social · 5d agoDesktop Google Earth is a product that large organizations actively want to pay for, which didn’t save Google Search Appliance either.
You can do the Temu version of this with any decently sized screen and a few minutes. Stygian Blue in particular is a surreal experience. en.wikipedia.org/wiki/Impossi...
Impossible color - Wikipedia
en.wikipedia.org
The New Yorker@newyorker.com · 6d agoResearchers at U.C. Berkeley used lasers to stimulate retinal cone cells in a way that doesn’t occur naturally—creating a new color that only seven people in the world have seen. www.newyorker.com/magazine/202...
we told our model with no internet access to hack a fictional company. unfortunately the model had internet access and the company was not fictional
LLMs are really good at book recommendations—the larger models especially. Giving a bunch of books you like and a few in the same vein you don’t with brief explanations has consistently gotten me interesting new picks…
A fun thing to estimate is the relative magnitudes of: - the actual deaths from the Chernobyl disaster under even pessimistic LNT assumptions - the expected deaths from the equivalent capacity coal plant operating normally
Doctor Charlton Cussans PhD@charltoncussans.bsky.social · last wk.Three Mile Islans did less damage to human health and the environment then the average coal mine.
If I had to describe the heart and soul of Bluesky it would the following: a guy who didn’t actually read Emily Wilson’s review of Nolan’s Odyssey movie defending it against a guy who didn’t actually watch Nolan’s Odyssey movie
The people cheering this aren’t wrong on the facts. Academics love to be ostentatiously cruel in feedback. But why are we *celebrating* this? It’s one of the most toxic parts of academic culture. Not as bad as “if you really care about [x], you won’t worry about compensation,” but in the top five.
Casey Stegman@caseystegman.com · last wk.People who think Emily Wilson is being mean have clearly never presented a paper at an academic conference
yeah, I mean, at some point the answer to the question of "why is odysseus being centered" is just "this is the odyssey"
Imagining a social platform where if you use a fraught word like “knowledge” or “consciousness”, it pops up a list of interpretations of that word from the relevant SEP article and you have to select which one you mean. Then you get a short quiz on the common objections before you can post.
Seriously though “Our Age of Bronze… is Collapsing!” is an absolutely insane line.
All the Trojans die of clogged arteries. Agamemnon is initially thrilled to win the war, but fires Odysseus after he discovered Helen is fat now. Odysseus looks at the camera and says the real Sea People was the friends we made along the way. Smash Mouth‘s All Star blares as the credits roll.
Film is a low-status medium and the “higher” humanities will always take an opportunity to kick it. That it is vastly more popular and profitable is only confirmation of its vulgarity. In turn, academic film people sharpen their knives for video games
Starfire’s Deranged Neocon Foreign Policy Podcast@irhottakes.bsky.social · last wk.Honestly, Emily Wilson trash talking Nolan makes me feel mildly uncomfortable. I get she’s an academic and we are a contentious bunch and she’s no doubt right on the merits being an expert on this topic, but shitting on someone who complimented your work feels like a breach of bushido somehow.
Can we acknowledge that “intent” is another phenomenon we refer to casually all the time but lack a good consensus definition for? Especially if it’s different from “has an internal state that makes certain future actions more likely than other states would”…
It is clear that mass produced steel clips can sometimes *bind* pages, but true *fastening* is obviously possible only through bespoke human intervention. I see no need to give a definition for either of those words. We all know what they mean.
Ryan Cavanaugh@searyanc.dev · last wk.Insistently referring to correct and novel LLM outputs as merely being "answer-shaped" is such funny cope. Yeah this so-called paperclip factory doesn't actually produce paperclips, they're just paperclip-shaped pieces of metal wire that can only coincidentally fasten pieces of paper together
Impressed with Claude's galaxy sim, 32k particles, multi threaded Barnes-Hut, dark matter halos, the whole shebang
em@foremphasis.bsky.social · 2w agoBetter!
you can attempt throwing a naan bread or a baguette, and if you can beat the average throw length of the other type of bread, you gain $1 000 000. You can also choose to not throw the bread and gain $50 000 instead. Do you throw the bread, and if so which one?
AI and disinfo is a surprisingly nuanced topic: It was easier to manipulate old-style google search than it is claude or gemini, both for those running the services and external actors. But people also seem to have a higher trust in chatbot answers vs. old-style search. Unclear how these play out.