You are in a hard takeoff. Buckle up.
>exponentially increasing slop + oil crisisyep we're in for a wild ride, I only hope israel survives
But I don't own any antiques.
>>110058112I don’t get it
>>110058186Neither does most of the world who has taken time to examine LLMs in any detail. Complete garbage. No intellience can emerge from it and it's fundemntal flaws of data loss and hallucination can never be resolved. Neither more trasining data or compute will help either.
^when prompting an LLM to make some typos to appear more authentic goes too far
>>110058144Of course you zionist faggots hate SI because it evens the playing field KYS
>>110058256It's not intelligent in that it can understand anything, but if you can guide it and know how to use it it becomes a very powerful coding tool. Anyone who's used it for serious, advanced projects knows this. The "one shot" web app demos you see on YouTube by third worlders who have no idea what they're working with are misrepresenting it.
>>110058308hehe i was being sarcastic, genius
One million shitty decomps and math "proofs" incoming
>>110058112Lmao is this the chart of projected token cost once SI is no longer subsidized by VC?
What's the source and what even is the chart about lol. What's an antique CR? Anthropic's mono is public or something?
>>110058256>No intellience can emerge from it and it's fundemntal flaws of data loss and hallucination can never be resolved. Neither more trasining data or compute will help either.The progress in the models since last few years suggest otherwise
>>110058395 (samefag)Oh it's from their blog.
>>110058256>he thinks LLMs are still GPT 3.5
>>110058311Worthlesd due to unfixable arbitrary data loss and random hallucination. Not fit for any purpose more than a toy/scam and never will be. Reality.
>>110058458GPT4 was just as retarded, and I doubt things have improved when there was no improvement to begin with.
op is a faggothere is the link https://www.anthropic.com/engineering/redesigning-source-control>In order to support this growth, we built Scion, an in-house version control system, and fully migrated our monorepo from git in five months. We also built and shipped a code review surface (called Antique), code search, a rich source control IAM system, novel ways of managing and landing large amounts of reviewed code, and more, all in that same time period. In this post, we’ll focus on how we built Scion specifically. It makes for an interesting case study: Claude’s ability to write code was both what put us in this situation in the first place and how we were able to quickly build a high-quality, scalable system to deal with it.
>>110058458If you knew anything about LLMs you would know data loss and hallucination are intrinsic to their design and cannot be fixed, nothinbvg in that has changed or ever will. People like you get filtered out as low IQ because you can;t even tell the difference between hallucination and real information due to low intelligence and educational failure. There is zero possibility of an LLM becoming intelligent, sentient and AGI or any other term. It is a lossy compression token probability predictor that generates random fictitious or gapped data results often and nothing more and that#s all LLMs will ever be. We ere told motre training data and compute might fix it. It did not and could never have. You have been conned in a giant investor fraud public relations campaign.
>>110058528cope
>>110058419>The progress in the models since last few years suggest otherwiseIf you think hallucination "has been solved", you're an ignorantfag.
>>110058528>If you knew anything about LLMs you would know data loss and hallucination are intrinsic to their design and cannot be fixed,based>>110058615>copeNo, you're a retard that doesn't understand the limitations of LLM, thus, most likely your piece of shit SaaS or software will soon produce misleading data that will end up in wrong decisions and loss of money or a company bankrupting.
>>110058112this only means there is zero chance the code in anthropic gets any chance of being reviewed.There will be very interesting bugs hidden inside and it's only a matter of time until they get exploited in shattering ways.
>>110058112All me, btw.
>>110058825you are wasting your breath. these vibetards think they won because they get a program that runs at all. they were never going to understand the code and they were never going to try either.
>>110058112funny because macos desktop notification just do not work, it's random if a notification shows or if the sound plays. It's pretty increadible. Also a chat yesterday had 90% of the latex work but places where it didn't
>>110058812I never said they are solved but if you think hallucinations haven't been largely mitigated then you are the ignorant one
>>110058112Oh shit they turned the linter on.
>>110058308>zionist faggots hate SIzionist faggots are the ones that own the SI companies, retard
>>110058528>>110058825These retards haven't realized LLMs have long since stopped deriving data strictly from the model itself and use any available data it has access to as sources of truth.
>>110058256I know more about how they work than you. The are intelligent in every meaningful sense of the word.
>have each agent constantly sharting industrial levels of vibe coded garbage into the same repo>"uhhhmm yikes, somehow we have 6 gorillion bytes of git traffic per second"Sounds like a skill issue
>>110061496just cuz u talk to ur ai gf all day long doesn't mean you "know" her better.
>>110058841Impossible to exploit when they rewrite a feature every 2 days
Will the takeoff resume in their revenue or not?
This time it's really over
