Olivia
Online
Suraj (al/acc)💭 🗣️🤺📡🔭🗝️✒️🎭☄
@Sur3_Altus
Building to life, a radically new way to connect meaningfully, & to make spaces come alive with personal art & stories
Live by: mm-aa
London
Joined June 2018
4.3K Following
659 Followers
5.5K Posts
Pinned Tweet
Introducing the DASPort, an AI-powered dynamic art & storytelling portal for homes & businesses.
Personal art. Evolving spaces. Intelligent ambience. Living walls that adapt to your moods, memories, stories, and life.
For spaces should come alive.
Explore more at deckoviz(.)com
at this point, consciousness definitions - & discussions - are like cars at an oprah show. everyone gets one, so it's all pretty meaningless. unless we add some serious epistemic rigour to these discussions, we might as well just be bouncing pebbles on the water.
Ted Chiang argues against octopus consciousness, apparently. Idk man.
I'm, um, genuinely uncertain about LLM consciousness, but this doesn't seem very serious.
In enterprises, we will start seeing the exact opposite of what we're seeing in consumers and AI, where people prefer human output over AI. Enterprises will prefer more & more AI output over human output for the exact same reasons: the efficiency of output, across dimensions.
In a new Stanford study, law professors by far preferred Gemini 2.5 Pro's responses over those written by their peers when they were unaware of who wrote the answers.
@teortaxesTex "so cheating with a more expensive model would have been fine? thanks for the update claudia"
Who to follow
the implication being that openai is more scientific progress driven, while anthropic is more financially driven. this checks out amply clearly from my vantage point. would make for a nice framing, if the line on the sand is drawn indeed.
When Mythos came out, my immediate thought was "if our models can prove 80-year-old theorems, surely they can find cyber vulnerabilities too." And they did.
I imagine the researchers there are thinking the same thought in reverse.
the irony being that they are being the exact thing that they are projecting these llms to be. it could not be more delicious. "in fighting stochastic parrots, beware you do not become one yourself," is something they skipped in school
Most people, including really accomplished people, don't have an accurate mental model of how LLMs operate (and why would they?)
You see this in wide beliefs that AI is just copying from known sources, or that it only produces average answers, or that it can't generate new ideas
"Attention is just a special case of <abstract math thing> so we generalized it by <neglecting the other 30 abstractions and conditions required for frontier architecture> and we found it performed <p hacking> compared to <naive baseline>"
and the Puzzle Maker is back in business after a short-ish hiatus. also, agi definition #14 when humans can't come up with a puzzle than ai's can't solve. so this role - in various labs - will be quite the interesting harbinger
We are now seeking a puzzle maker to help us create puzzles that LLMs can't yet solve.
much greater roi in building out a skill stack with 4-5 skills - instead of going from 95pct to 99pct with diminishing returns, spending that time in adding another skill or two and getting it up to 90ptile
In most professions, when you hit the 90th percentile in technical skill, the best use of your time is getting 90th percentile soft skills
the best barbers are therapists
the best photographers are comedians
the best engineers are iconoclastic cult leaders that inspire their team with a vision nobody else can see, and even fewer can beleive
stop maxing out your CAD skills, and start dreaming bigger!
how it feels to be a hardware engineer visiting shenzhen
maybe, just maybe, treating the dating & relational landscape as a marketplace is the exact reason why it's become so bleak & so many people feel unexcited about it, many even checked out. the commodification & transactionalism writ large in a landscape where it does harm
"dating out of your league" isn't real. dating is a marketplace. it's called price discovery. your value is whatever clears the market
question: if attaining the perfect balance - of when to push back, & how much to - was impossible (for now), would you rather your model lean more towards being more agreeable than the ideal, or more towards being less agreeable than the ideal?
After using Opus 4.8 extensively I assume there is some panic internally at Anthropic as it seems they are defaulting to cheap post-training tricks to get the models to "push" back.
Which means that the biggest failure mode of wrong token paths out of which the models lack the meta rationality to pull themselves out is a huge and non trivial issue.
in times of great transitions and disruptions, philosopher's gonna rip. philosopher, the sexiest job of 2030's?
One of the funniest sentences I’ve read in a while
claude, what's your drama setting?
claude: 30 - I am pretty chill, see
"some kind of computational infidelity" (Opus 4.6)
“wow you really must love mango orange”
words have a remarkable power to gigafry people's brains - we didn't need llms to learn this. people are willing to die for words, to kill for words, to love boundlessly for words, to ascend it all. ai psychosis was the most expectable thing to come out of post-turing-test models
this is an interesting point in the new ted chiang piece – no one really claims that alphafold is conscious, or that sora or midjourney or dall-e are conscious
@mickeymuldoon Haha, hopefully, not for too long though, for my sake and its.
When Claude messes up, I call it a silly goose. When ChatGPT messes up, I call it a clanker. I am starting to see the first signs of "AI-ism" in my responses to these models, and I'm sure I'm not the only one. This is the new racism?
knowing when to disagree, and how to disagree, and how not to, to increase long-term agreement & congruence: a core life skill.
A common, well-understood sales tactic is "Don't do this if isn't the right deal for you: here are some reasons it might not be."
It works even if you say "By the way this is a sales tactic. It should demonstrate to you that I actually care about your happiness." after it.
we've got these magical learning machines that can teach us anything at any pace, at any level, for billions, and you have the gall to complain about fabricated & hallucinated water issues? the audacity & lack of imagination & faith in humanity is astounding.
i just love that i can zoom in on new scientific fields at whatever resolution i want now. i can choose a random area, say give me the two pager on what this area has learned. if i like it ask for the 10 page. then clusters of papers, then individual papers. so natural and good
it's official. agi is here. and it's undeniable
for real, though, we need stronger language-based benchmarks, because agi will emerge through that vector, not math and economy-based evals. humor in language, self-awareness through language, creativity expressed in language, etc.
oh my god
@omniderek peak bubble behaviour. sell it all
Last Seen Users on Sotwe
Most Popular Users
Elon Musk 
@elonmusk
240.1M followers
Barack Obama
@barackobama
119.3M followers
Donald J. Trump
@realdonaldtrump
111.6M followers
Cristiano Ronaldo
@cristiano
108.7M followers
Narendra Modi
@narendramodi
106.9M followers
Rihanna
@rihanna
97.2M followers
NASA
@nasa
92.1M followers
Justin Bieber
@justinbieber
90.5M followers
KATY PERRY
@katyperry
86.7M followers
Taylor Swift
@taylorswift13
80.5M followers
Lady Gaga
@ladygaga
72.1M followers
Kim Kardashian
@kimkardashian
69.3M followers
YouTube
@youtube
68.6M followers
Virat Kohli
@imvkohli
68.4M followers
Bill Gates
@billgates
63.4M followers
The Ellen Show
@theellenshow
62.5M followers
CNN
@cnn
61.9M followers
Neymar Jr
@neymarjr
60.9M followers
X
@x
60.9M followers
CNN Breaking News
@cnnbrk
59.9M followers