Sabitlenmiş Tweet
me & @asiyayyay had a random thought:
what if someone else could peel the orange for you?
so we built lazySantara
check it out:
lazysantara.com
English
Shivam Verma🎴
578 posts
Shivam Verma🎴
@sheeeevam
building @lazySantara https://t.co/7khGtrEieV
Bangalore, India Katılım Şubat 2018
421 Takip Edilen82 Takipçiler
Shivam Verma🎴 retweetledi
tried something new
wanted to write about something beyond tech
open.substack.com/pub/seawam/p/s…
English
Shivam Verma🎴 retweetledi
i made a chrome extension that removes the AI slop from my linkedin feed
English
Shivam Verma🎴 retweetledi
this was made with ChatGPT Image 2.0.
we are so cooked 😭
OpenAI@OpenAI
Introducing ChatGPT Images 2.0 A state-of-the-art image model that can take on complex visual tasks and produce precise, immediately usable visuals, with sharper editing, richer layouts, and thinking-level intelligence. Video made with ChatGPT Images
English
Shivam Verma🎴 retweetledi
pretty pretty useful
Linus ✦ Ekenstam@LinusEkenstam
Huawei camera AI now recommends poses before you click the shutter button
English
one of the coolest pitches I have ever seen
Yaz@yazcal
I keep coming back to this. in 1666, leibniz, 20 years old, wrote de casibus perplexis in jure. perplexing cases in law. around the same time, he was already imagining a formal calculus of reasoning. his provocation was simple: when two people disagree, stop arguing, pick up pens, and say calculemus. “let us calculate.” lawsuits, he thought, should end the way arithmetic does. 360 years later, still trying. and even then, the dream was older than he was. hammurabi, 1754 BCE: šumma awīlum. “if a man.” 282 times. condition, consequence, exception. pure if-then. the ancestor of every codebase since. then aristotle pushed back. nicomachean ethics, book V. the builders of lesbos used a lead ruler that bent to fit irregular stone. equity works the same way. law has to bend to the shape of the case. that is the whole argument, basically. rules vs standards. code vs judgment. straight edge vs lead ruler. it never ended. hart called it “open texture.” holmes put it more cleanly: “the life of the law has not been logic; it has been experience.” every serious attempt to compile law runs into the ruler. the ruler always wins. then LLMs happened. for the first time in almost 4,000 years, we built something that seems to bend like the ruler. “reasonable.” “good faith.” “material adverse effect.” models read those the way judges often do: probabilistically. contextually. by analogy. nothing got solved. the rule-vs-standard distinction did not collapse. it shifted. interpretation now lives in model weights we cannot read. programmability became interpretability. alignment is jurisprudence with worse vocabulary. so the question is no longer whether law can be made executable. when a model decides what “reasonable” means, who wrote the law? quis custodiet ipsas machinas. we're building programmablelaw.com because quis custodiet has an answer: whoever has to live with the machine. right now, the lab sells you a machine it calls phd-level, keeps the steering wheel, and calls it probabilistic when it drives your car off the road. we're moving the wheel.
English
@d4rsh_tw cause the map we grew up with (mercator, 1569) was made for navigation, not true size
it distorts areas near the poles, see this thetruesize.com
English
Your one stop destination to LLM benchmarking and evaluation:
artificialanalysis.ai/evaluations
English