https://www.youtube.com/watch?v=Cmoh-yR-usAThoughts on this foid? Basically she goes over how she has a bunch of agents for 24/7 coding and pushes like 1000 PRs a month.My thoughts:On one hand, she is not basing any of her conclusions on how the technology works. I think it would make sense to think about how LLMs actually work, and draw conclusions on how to use them from there. Instead, she's basing her method on trial and error + "developing trust in the agents".At one point she said that the agents leave stupid comments in the code so she banned code comments, but she has a series of rules in place so they can't output shitty code. This is an obvious misunderstanding of what shitty code is, because these days there's a new kind of bad code which is great on paper but way too long for its purpose, bugged, and makes no sense. Some devs always did that, but it exploded with AI.I can see how code quality could be less important if only agents work on the codebase, but I wonder if that also means the problems will compound.On the other hand, she's obviously smart, and she has a point. I can see how it can make sense to do things the way she does and the results could be good or at least acceptable + much faster and cheaper. I would not be dismissive of the potential of this.So basically foidslop and AIslop but has a point.Then there's another problem, the obvious problem of people losing their jobs. Maybe AI will create more jobs and it will be even better in the end, this is in fact probably the case, but I still hate it when people like this talk about corps becoming "nimble" and firing everyone and they don't care about actual people suffering, like its not even a concern to them.
>>109663957>another simp thread on /g/
>>109664256are you retarded
She looks like that pink plushy thing that was famous years ago
>>109663957
>>109664260>watches an hour long video of a literal who roastie vibecoder>writes a blog post about said roastie on gee>is offended when called out for being a simp
How is xhe reviewing 1000 PRs a month?
>>109664359i dont care about this gook whore. i am saying that she's pushing foidslop. but this is undeniably relevant. and she is somehow very highly paid with key positions in top tech companies.>>109664486she's not. she's saying that you can slowly build trust in the agents and stop reading the code basically
>>109663957> code quality could be less important if only agents work on the codebaseConcept of a tech debt still exist, throwing "SlopBot, fix that thing" at it doesn't make it disappear.
>>109663957>Thoughts on this foid?
>>109664504I know, but I can see how, maybe, priorities on what good code is could change.
>>109664519I see your point, but I don't think it's the case.LLMs are not some kind of new intelligence, they are trained on existing code, mostly made by people - for people. So, both as input and output, best performing results are likely imitations of the human-produced code.
>>109664559Good point ngl
>>109663957your post sure looks like aislopi dont rly care about your foid
>>109663957If she's the bottleneck to verify if the code is correct, how does the agent verify it itself?
>>109663957https://paste.rs/PuwH3.mdhttps://www.aihackforfreedom.org/these pieces of shit are just deep state karens and teachers union deadshits, people tied to USAID, every fed or bureau cunt that 4chan collectively despises is behind this, it's likely whatever agentic capability is developed here would be used domestically in the US by these fuckers, probably in time for november.I'm not joking. they came into a very exclusive discord full of xooglers working on their own AI projects advertising to some very skilled developers that they should participate, they're looking to do something big and I witnessed them trying to recruit.they implied the tools would be used by people "aiming to topple a regime"I shit you not.
>>109663957>pushes like 1000 PRs a month.Is she paying someone to review them?
>>109663957Why won't he at least try to put on a female voice if he's tryin to be a gril
This is just folklore system where engineers are wasting token running a complex system that can become redundant the moment AI labs re-calibrate their models, or when the organizations have to switch to a different vendor due to licensing / budget. It's like early day agile/scrum where no one can reliably verify the effectiveness of said process.
how do people afford to do that? is she rich? $20 codex sub runs out almost instantly when trying to do Real Work
>>109668371Her company paid for it - I heard Meta clamped down the individual quota because of "abuses" like that.
>>109668384how are they not fired? if you just let a machine crank out a bunch of shit you have no idea what it does in another job, you'd be insta-firedi guess i'm too retarded for tech. irl there is consequences for mistakes. what am i not getting here?
>>109663957AI agents are useful but this "le 24/7" psychosis is worthless trash coming from idiots who are more obsessed with their "productivity" than they are concerned with actually producing anything useful that real people outside the same obsession bubble actually use.
>>109668390Probably because they were told to find ways to leverage AI
>>109665953You fucking imbecile, I hate it when unintelligent people think everything is AI.If you think everything is AI it means you can't tell AI slop apart from what people write, meaning, you're literally no better than the normies who are perfectly fine with AI slop because it makes sense to them.While writing style is more obvious, and I can see how someone with limited cognitive abilities may think "this is well written and long, must be AI", in reality the main thing about AI slop text is there's no thoughts behind it. It's just trying to write something similar to other texts on the topic, whereas a person's writings reflect someone's thoughts. If you can't tell the difference you're just a normie.
>>109667942Her argument is that you can build trust in the agents and, once you trust them enough, you stop checking the code.>>109668107agents review it kek
>>109668390Just watch the video. But this would be my take:First off, she is a foid in tech which is a huge advantage. However she must be smart because she still definitely made it farther than most foids.But also and most importantly, just notice how good she's at talking. From hearing her, you'd think that she basically figured out just the right way to set up AI so it'll do everything right and be equivalent to a thousand software devs. The average person who has no inner thoughts and only responds to stimuli, will believe exactly that after listening to her. Being able to promote herself like this is why she's rich.As to what's her job after setting that up, maybe she does nothing all day, and maybe she's really busy supervising this setup and improving on it. We can't tell, and whether she does any work or not has no effect on her employment, what her boss believes does. And from how good she's at talking, her boss probably believes she's 100% needed in order to keep this running properly and constantly improve. They're not going to say "ok you already set that up now you're fired" because the whole point is that she's supposed to be constantly refining it and improving it, and without her they wouldn't have the best possible technology.
>>109668390techtards be like "move fast and break things" they actually live by this and think it's good which is why theres a breach every other month or some other catastrophic windows or cloudflare outage that wipes out half the internet or etc etc etcthey have no sense whatsoever of caution or accountability, or even the most basic sense of care for anything and i mean why would you when you make exorbitant amounts of money, all but directly run the world, and are all but immune from consequences except in the most egregious possible of cases where they make an example of someone once a decade (and their punishment isn't even that bad), like sam bankman, who was really more of a finance pariah too than tech, all that matters is pushing up the kpis and thus profitsso it doesn't matter if the people working for them is the most negligent fuck in the world, generally. Everyone in tech makes 500k a year can can absorb the loss of a job like nothing and jump ship to some other company in a couple months anyway after spending time vacationing so they have content for their blog that they work on while they're supposed to be workingit's a farce industry
>>109664346lolol
code quality becomes less relevant as ai improves, assuming its algorithm is optimal, as the ai is the only one reading it. Infact the only reason programming languages and assembly language exist is to make it easier for humans to code, the compiler does its best to turn it into binary. AI in a few years will be writing purely in binary as that would result in the most optimal software. So there's no reason to have human developers anymore, the few CS jobs left will be to monitor the systems basically an even more automated version of what she is doing.
>>109663957How much money has this cost and how much money has she earned ? I mean AItards are completely braindead but I really don't understand how you cannot see that this is essentially saying that you are doing and contributing zero things, which obviously means whatever you do I can do, rendering all software made this way just a function of how many tokens you or I can afford.>but muh AI generated promptsYeah, might as well start believing in astrology if you think writing thanks three times will make your Claude subscription better than mine
