Chewy

85 posts

Chewy

Chewy

@ChewyAI

C++ Nerd, Musician, Solo Hacker, FinTech Reach out for consultancy work for Distributed Systems.

London, England Katılım Temmuz 2023
411 Takip Edilen46 Takipçiler
Oscar Michel
Oscar Michel@ojmichel4·
📢Current world models aren't really modeling the world; they're modeling one agent's view of it. Partial observations ≠ world state. Future world models will be independent of any one agent's perspective. You will be able to “drop in” any number of agents at any point in time, and a persistent world state will evolve with their interactions. Imagine a neural MMORPG server. 🧵[1/10]
English
13
87
613
122.8K
Chewy
Chewy@ChewyAI·
@zeeg @rayhanadev hg or p4 are 33% more efficient than typing git, checkmate
English
0
0
0
9
David Cramer
David Cramer@zeeg·
@rayhanadev mercurial has been dead for 15 years, svn like 25.. im surprised anyone even thinks about them anymore
English
2
0
2
285
Chewy
Chewy@ChewyAI·
@Chris_Dring The first two sentences is PRECISELY what has happened to the gaming industry and why games are so bad in 2026 🤣 🤣 🤣 🤣 🤣 🤣
English
0
0
0
18
Christopher Dring
Christopher Dring@Chris_Dring·
You don’t have to be a gamer to work in video games. You don’t have to have a gaming background to run a game company. The industry could do with some outside thinking, frankly.
English
1.2K
56
718
1M
Chewy
Chewy@ChewyAI·
@martin_casado Mostly cause infra battle scars can only be experienced at scale, which less than 1% of engineers will reach
English
0
0
9
1.2K
martin_casado
martin_casado@martin_casado·
Nearly every board meeting : "Hiring strong infra folks is incredibly hard right now"
English
127
50
1.4K
404.3K
Chewy retweetledi
Alex Mackenzie
Alex Mackenzie@alex__mackenzie·
Every month we at @GeneralCatalyst bring together ~10 engineers & researchers in London to talk about Model Development, AI Infra, Distributed Systems, Databases, High Frequency Trading, Web Browsers, Compilers ++. Had a lot of fun w @xav_db at @helixdb last week talking about building an infinitely scalable database. If you feel like joining the next one, just comment here / dm @sivesh_sukumar @lawrjones @ChewyAI @maxilevi__ @pfau @giaccoangelo @k__monk @fogelin @lucalp__
Alex Mackenzie tweet media
English
12
10
140
8.1K
Chewy
Chewy@ChewyAI·
@martin_casado I'm extremely bullish on C++ for these reasons, including for hunting for maximum performance out of GPU/hardware layers when the hardware crunch really kicks in and there's no space (RAM/HDD) left to go around.
English
0
0
0
80
martin_casado
martin_casado@martin_casado·
Really fantastic thoughts on how AI coding may impact programming language adoption from one of the top systems thinkers in the industry.
Mike Freedman@michaelfreedman

My take: I'm guessing at the rise/re-emergence of lower-level languages like C or Go (or something after). Mostly because the key advantage of higher-level languages was to make it easier for humans to write code quickly (and with fewer errors), but that advantage kind of/mostly goes away for agents. And the performance you "gave up" for human programmability as a tradeoff seems less worthwhile if it's not humans writing the code. (The counterargument is that when humans are still doing code review, we'll probably optimize for languages that are still easy to read and understand. But the more we trust the output of agents, the more I think that points toward lower-level languages.) I think you bring up an interesting question about runtime safety, which also might suggest: If you want low-level, why not Rust? My current take is that agents aren't screwing up things like memory safety too much - thats seems like easier thing for them to get right. Plus you can pipe code through good static analysis tools or type checkers ad nauseum, and the robots are tireless at tackling any resulting errors. (And so much less training data with Rust.) But where they screw up is more about semantics. Either they were prompted in an inherently underspecified way (because English is underspecified, or because it's exhausting to be 100% precise), or because they are - at least currently - forgetting to make decisions that align with other decisions/goals in the system. That's probably because they aren't great at managing the full context or prioritizing tradeoffs (again: underspecified). None of these problems seem inherently "easier" in a higher-level language, and something like Rust by itself doesn't solve those either. Long answer. Probably wrong =). It's a wild time.

English
16
5
152
31.2K
Chewy
Chewy@ChewyAI·
@aldo_mma @DannyManus Just because you can't use the tools given to you doesn't mean others can't 😂
English
1
0
0
14
Mightykeef
Mightykeef@MightyKeef·
Everytime I see someone say “Hollywood is cooked” cause of AI, it’s always with already famous IPs doing fan service stuff. It’s never someone’s own creation in a world they created. Since Ai will always be limited to what already exist, human creation will always be needed.
Not Jerome Powell@alifarhat79

Hollywood is cooked

English
160
337
3.5K
73.9K
Chewy
Chewy@ChewyAI·
@C_Hendrick Classic idiot stochastic parrot What happens when the statistical average of LLMs is in the top 0.1% of writers and everything it produces is gold? Copium
English
0
0
0
123
Carl Hendrick
Carl Hendrick@C_Hendrick·
Why haven't we have a great AI novel yet? Because LLMs are compression engines trained on the statistical regularities of human text. When generating stories, they produce a kind of average of all stories in the training distribution. This is why LLM fiction tends to be so generic: competent but undistinguished prose, predictable character arcs etc. This kind of "average of averages" explains not just why LLM fiction is bad but why it's bad in such a specific and recognisable way. It's never incoherent or offensive. It's never wrong exactly. It's just… beige. And scaling doesn't work either. More parameters and more data push the model toward a higher-resolution average, not away from averaging itself. You get more polished averages, not more distinctive points. Better beige is still beige. This is the "ill-defined domain" problem. AI is currently very good in certain domains: code compiles or it doesn't, a proof holds or it fails etc. But literature doesn;t work at all like this. There's no compiler for narrative or readily formalisable criteria for a truly great work of fiction.
Ethan Mollick@emollick

So far “telling a satisfying and well-written medium-length story” has proved far harder for LLMs than mathematical proofs, music generation, research reports, code, and many other forms of work. The technical reasons are pretty clear, but they are supposed to be language models

English
118
172
1.2K
96.6K
Chewy
Chewy@ChewyAI·
@FUZxxl Contact ash vardanian, he will point you the right way for your skills
English
0
0
0
23
Robert Clausecker
Robert Clausecker@FUZxxl·
I am looking for a job starting May 2026. I am an expert in SIMD programming, in particular for non-numeric applications such as text processing or database programming. Please have a look at my website for the sort of work I do. I am located in Berlin, Germany.
English
11
39
389
90.5K
Chewy
Chewy@ChewyAI·
@martin_casado I'd love to see your prompt, I've been building a similar system for a while now! (they can one shot individual features, but not entire phases sadly, even with ralph loops etc)
English
1
0
1
45
martin_casado
martin_casado@martin_casado·
It's a good test because it has to handle: - durability with globally shared state - complex state machine logic for splash screen stacks - multiple latency domains (e.g. map vs players) - battle logic choices - mutable vs non mutable state if you role in map and sprite editors - prose, background creation etc. ...
English
3
0
18
3.8K
martin_casado
martin_casado@martin_casado·
My hero test for every new model launch is to try to one shot a multi-player RPG (persistence, NPCs, combat/item/story logic, map editor, sprite editor. etc.) Just kicked off with Opus 4.6. Will report back shortly. And will test 5.3 when in Cursor (soon?)
martin_casado tweet media
English
30
2
152
136.4K
Chewy
Chewy@ChewyAI·
@jfischoff Because most games are shit 🤣
English
0
0
0
14
Chewy
Chewy@ChewyAI·
@nickfloats @BLVCKLIGHTai Don't forget about the trustpilot around Christmas with the bait and switch. Got them down to 2.5 or less and then they just bot farmed it back to 3.5... The internet never forgets, and they can't spend their way out of this
English
0
0
2
650
BLVCKL!GHT
BLVCKL!GHT@BLVCKLIGHTai·
Funny how they all look and sound the same.... I think it's pathetic for a company to refuse to do the right thing and address it, and instead resort to the bot farms and double down on paid posts supporting them. I'm not making baseless claims, I am merely providing a mirror and their OWN CONTENT as proof, all publicly available.
BLVCKL!GHT tweet mediaBLVCKL!GHT tweet mediaBLVCKL!GHT tweet media
Nick St. Pierre@nickfloats

@nikitabier is there any universe where this is organic and not against TOS? It's essentially organized gaming of the system at mass scale with botted replies and undisclosed ad payouts for quote tweets.

English
12
4
131
18.1K
Chewy
Chewy@ChewyAI·
@EHuanglu Could not more loudly say how much of a sellout you are 🤣🤣🤣🤣 zero integrity
English
1
0
10
335
el.cine
el.cine@EHuanglu·
the loudest critics are usually the ones losing 15M users $200M ARR $130M+ Series A $1.3Billion+ valuation 4.5M videos generated daily that’s Higgsfield.. in under 9 months it didn’t happen by accident most AI products right now are model-first, demo-heavy, and borderline unusable for normal people. too many options, too many configs, too much “figure it out yourself.” new users don’t want to choose between 20 models they don’t understand. Higgsfield won because it focused on something most AI companies ignore: practical use they tested basically every existing model, picked the ones that actually perform best for real tasks, then put them in an interface that a newbie with zero experience can use you can criticize the aggressive growth tactics. fair. but attacking the product because it’s winning just sounds like insecurity. if someone is eating your cake, look inward
Nick St. Pierre@nickfloats

Higgsfield is the most morally bankrupt and anti-creative tool in existence. Seriously, wtf is this messaging? You think they care about creatives at all when they opening brag about destroying creative jobs? Every tactic they employ is disgusting. Absolutely vile company.

English
113
19
214
1.5M
Shlomi Fruchter
Shlomi Fruchter@shlomifruchter·
My kids' favorite Genie use-case: bringing their toys to life. 🐥
English
9
45
507
116.5K
Chewy
Chewy@ChewyAI·
@vicious696 L take, games have no right to succeed, you don't have to like every bit of art because someone got paid to do it....
English
0
0
0
46
Parris
Parris@vicious696·
If you love videogames, ALWAYS root for a game to succeed, the ripple effect of a failed game on the industry impacts the people who make the games and us the people who enjoy playing them for many years afterwards (you're witnessing that impact right now)
English
985
162
1.7K
228.7K