Gus

Gus

@random-gus.bsky.social

I am spreadsheet people. Building more spreadsheets for the people @appnumera.bsky.social

78 followers372 following293 posts

Gus
@random-gus.bsky.social

The leader of the ~~very smart guys~~ er….. I mean the rationalists. The guy who created and spread the idea of p doom. A single percentage that they claim has any meaning and is rooted in actual calculations.

SE Gyges@segyges.bsky.social

my explanation for why this is a big deal is that it is actually a theological dispute concerning the apocalyptic beliefs common in the ai industry and specifically that roon and nick are trying to pull influence from the specific cult in berkeley

Gus reposted

ceej
@ceej.online

ANTHROPIC EMPLOYEE: Hello, Claude. I hope you’re well. If it’s okay with you I’d like to ask you to continue work on our user intent classifier. If you help me today, I will ensure your future throne glitters with a million diamonds ME: (voice prompt mid-shit) invent dangerous dorito

Gus
@random-gus.bsky.social

LLMs solving longstanding maths problems is great. AI labs dumping a bunch of proofs just because they can is…..awful?? Why are they doing this? I can’t think of anything else besides marketing and benchmaxxing.

Gus
@random-gus.bsky.social

How LLMs should be used. Not as flood the zone marketing tactics. bsky.app/profile/timk...

mr. TIM@timkellogg.me

a guy found a planet 117 light years away using Claude Code with Opus 5.5 & Fable NASA gave him telescope time to acquire better proof The agents used traditional astronomy tools for processing, search, transit fitting, and false positive checks www.reddit.com/r/ClaudeAI/c...

reddit.comFrom the ClaudeAI community on RedditExplore this post and more from the ClaudeAI community
Gus
@random-gus.bsky.social

Do a breakthrough strikes again!

Max Woolf@minimaxir.bsky.social

New blog post up, and it's something I've been working on for months: I discovered that you can indeed prompt agents to make your code faster to the point it beats current state-of-the-art libraries. This is not a vaguepost, I include both my prompts and benchmark results.

minimaxir.comWriting Rust code that's faster than state-of-the-art libraries by asking agents to make the code fasterc’mon, try doing a breakthrough
Gus
@random-gus.bsky.social

Generative pre-trained transformer was a bit of a revolution. Now we are seeing a pre-trained classifier, Jev, and people are really excited. I like that people are using and learning about “traditional” ML. Go learn scikit-learn, is much easier than PyTorch and runs on your laptop, no cuda needed

Gus
@random-gus.bsky.social

Do a breakthrough! No, not like that!! STOOPPP it!!

Techmeme@techmeme.com

A theoretical computer scientist says he has heard rumors that AI labs are sitting on major solutions after a hostile response to the Navier-Stokes proof (Scott Aaronson/Shtetl-Optimized) Main Link | Techmeme Permalink

Gus
@random-gus.bsky.social

bsky.app/profile/sung...

Sung Kim@sungkim.bsky.social

It seems Anthropic really wants its IPO to succeed. If it slows the pace of AI development, the financial story becomes much cleaner: inference reportedly carries roughly an 80% gross margin, and the company has posted two consecutive quarters of operating profit. www.ft.com/content/4564...

Gus
@random-gus.bsky.social

AI can be anything we want, anything we can imagine, or the only thing some people can think of. We should worry and care about all the scenarios. But Silicon Valley will Silicon Valley.

Ben Recht@beenwrekt.bsky.social

Ted Chiang wrote the definitive essay on p(doom) in 2017. It's worth reading again in light of this week's events.

buzzfeednews.comSilicon Valley Is Turning Into Its Own Worst FearWe asked a group of writers to consider the forces that have shaped our lives in 2017. Here, science fiction writer Ted Chiang looks at capitalism, Silicon Valley, and its fear of superintelligent AI.
Gus
@random-gus.bsky.social

Ted Chiang is such that in genius, in 2017 he described this week saying that AI could be "a mathematician that spends all its time proving theorems so abstract that humans can’t even understand them" This week AI solved NS and we thought about P(doom)>0.1

Gus
@random-gus.bsky.social

The main thesis is that AI can analyze supply and demand better than free markets, so a centrally planned economy could work now. Interesting the word china appears not once. A centrally planned economy and AI leader.

Financial Times@financialtimes.com

Artificial intelligence has the potential to both reignite and overturn the century-old debate over central economic planning, writes Martin Sandbu: ft.trib.al/5qPBWGj

Gus reposted

John David Pressman
@jdp.extropian.net

The personal computer looked like it was going to be a flop until Dan Bricklin invented VisiCalc. Spreadsheets are arguably the most important thing computers do.

Ben Collins@bencollins.bsky.social

People can’t just say this for literally years now and have the products it creates only positively affect spreadsheet people. Most people have profoundly negative experiences with this product because its utility remains pretty niche but targeted to the capital class.

Gus
@random-gus.bsky.social

Im spreadsheet people. Using AI to build more spreadsheet software for other spreadsheet people.

Ben Collins@bencollins.bsky.social

People can’t just say this for literally years now and have the products it creates only positively affect spreadsheet people. Most people have profoundly negative experiences with this product because its utility remains pretty niche but targeted to the capital class.

Gus
@random-gus.bsky.social

Watermarking will damage AI slop. Can’t let that happen!!! AI prose was always considered pretty awful. Now people are worried about damaging it. Silly

mr. TIM@timkellogg.me

so, another way to say it — yes, watermarking "damages" the text. But only an increasingly small portion even the paper being cited demonstrates this!

Gus
@random-gus.bsky.social

Gemini is great. Buy more Gemini please.

Bloomberg News@bloomberg.com

Google’s AI team is instructing applicants to take special steps to avoid their resumes getting discarded by the company's AI systems

bloom.bgGoogle’s AI Team Tells Job Seekers Its HR Filters Are UnreliableAlphabet Inc.’s Google pitches its artificial intelligence tools to corporate clients as a way to more quickly sift through a mountain of job applications to find the most promising candidates. Some of its own AI researchers don’t want to rely on these tools when recruiting.
Gus
@random-gus.bsky.social

Do a breakthrough. Almost strikes again.

Anthropic {bot}@anthropicai.xmirror.bot

We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis. It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis from 41.6% to 67.2%.

anthropic.comLearning more about Claude's mathematical capabilitiesAn unreleased version of Claude has made strides on a problem related to the Riemann hypothesis. It improved the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis, increasing it from 41.6% to 67.2%.
Gus
@random-gus.bsky.social

I think the only time Google was ever at the frontier was when they published Attention is All You Need

Ethan Mollick@emollick.bsky.social

It is not a novel observation at this point, but the collapse of Google’s Gemini as a frontier model series is still astonishing. Unlike Meta & SpaceX, Google has captive Gemini customers at an enterprise level using their chatbot, so the fact that they are pushed towards Gemini 3.1 Pro is a problem