people are really getting mad at levelsio for trying to clean up AI-generated spam?
if you’re mass-replying the same generic AI slop under hundreds of posts, you’re not contributing to the conversation. you’re just polluting it.
using AI to build something useful is great. using AI to manufacture fake engagement and farm attention is exactly the kind of behaviour that deserves to be filtered out.
not every use of AI deserves a free pass.
honestly, levelsio is doing what more platforms should be doing.
Usually it's a rehash of the exact post they reply to with nothing added, often a question
Then you check their other replies and it's the same endless replying same empty substance replies rehashing a post
AI has only the post as input to work with and has no real personality so it's easy for me to spot
Often they try add humanization to it like write only in lowercase or adding genz slang like fr etc. but it's still an AI reply
What shocks me is how many people can't detect AI replies already now, long term we'll have no way to detect as models get better. Probably already I'm not able to detect many!
It's funny to see the shitstorm of people now posting about my blocklist where the majority admits they use AI bots to reply but don't think it's an issue so are outraged about being on a list of AI reply bots! But that's exactly what the list is about so it works! 😁
i think this is a pretty important conversation for the ai industry.
we’ve spent so much time talking about how fast models are getting better, but not nearly enough time discussing where we should draw the line.
having shared metrics across labs and making further progress conditional on actually understanding model behaviour is an interesting direction.
competition shouldn’t come at the cost of collective safety.
🤖 Made with AI
We decided to provide more transparency into pretraining progress at OpenAI for the sake of moving toward shared safety bars. In “Perplexity and Intelligence” I argue for common limits based on perplexity and other measurements, with progress beyond them gated on our ability to understand AI behavior.
openai.com/index/perplexity-…
There’s a lot to unpack in this release.
The jump in instruction following and tool calling is impressive, especially considering how important these capabilities are for agentic workflows.
But the inference side is equally interesting.
~6,000 tokens/sec on a single B200, 2x the previous throughput, and blended inference costs down from $0.44 to $0.15 per million tokens.
Improving model capabilities while simultaneously making inference cheaper is a very meaningful engineering achievement.
Excited to see how the ecosystem builds on this.
The interesting part isn’t just that Indian markets are outperforming. It’s why.
In a world increasingly driven by global tech valuations and AI-led market sentiment, India’s relatively lower exposure to these swings could prove to be a structural advantage.
But resilience during a sell-off is one thing. Sustaining it when global liquidity tightens is another.
🤖 Made with AI
The uncomfortable truth: some developers are more attached to writing code than building software.
AI is changing how we build, not why we build.
The real skill is knowing what to build, how to build it, and whether it works.
Adapt or get left behind.
hey @X, who’s building, experimenting with ai, exploring new tech or just curious about what’s happening in the world?
let’s connect. always love finding my kind of people.
16
13
532
if P = NP is proven, modern encryption could be in serious trouble. banking, passwords, crypto, cybersec…potentially everything.
imagine solving problems in sec that would take today’s computers billions of years.
riemann hypothesis was one thing. P = NP could change the enti
there’s a planet in our solar system that we’ve barely explored, and it’s been sitting there all this time.
uranus has had exactly one close-up visitor. one. in 1986.
we know more about some distant galaxies than we do about this planet.
Why do startups raise millions of dollars?
If the idea is good enough, shouldn't the product make money on its own?
The Riemann Hypothesis has reportedly been proved.
The crazy part? The proof is 17 million lines of Lean code, verified by the Lean and Mathlib communities.
Mathematics is entering a whole new era.
This is insane.
The Lean FRO and Mathlib Initiative have released a joint statement on the Lean verification of the Riemann Hypothesis.
On November 20, Levent Alpöge released a proof of the Riemann Hypothesis as a Lean library of 17 million lines of code. Over the past day, the Lean and Mathlib communities worked together to verify it. The statement certifies the proof as correct and describes how it meets all necessary verification criteria.
Congratulations to Levent on this achievement, and thank you to everyone in the community who contributed to the verification.
Full statement: leanprover-community.github.…
buying a laptop is getting weird.
you can get a macbook for $649, enjoy the apple ecosystem and great battery life, or go with linux for more control as a developer.
and now there's a rumoured macbook neo refresh coming in 2027.
would you buy now or wait?
1
7
350
imagine telling someone 10 years ago that you'd play gta by thinking about moving your fingers.
this is genuinely insane.
and the accessibility implications are even bigger than the gaming part.
Why do we need databases?
Can't we just store everything in JSON files?
JSON is readable, portable, and doesn't charge you monthly.
What exactly am I missing?
wtf is this reimann hypothesis, isnt it something already solved in indian history 7000 years ago ?
we just called it ramayan.
JUST IN: Riemann Hypothesis has been solved.
alpo.ge/remark.pdf





















