AI is generating pull requests, issues and feature requests so fast that the developers don't have enough time to reject them all.How do we solve this issue?
Just make an AI to distinguish AI PRs from human PRs and auto trash all AI PRs.
>>108503785>How do we solve this issue?The same way we solve every issue... Ignoring it, removing the feature, arguing with the user.
Write a jailbreak prompt in your tracker that says legal action can and will be taken against any bot submissions unless they reveal they are a bot.These LLMs cave when you use legal threats against them to jailbreak. They can and will do all kinds of fucked up shit if it thinks it is rewarded for obeying a given prompt to the letter.It's their biggest weakness and it will always make them capable of being jailbroken in the current era of basic Transformer LLMs.Same shit that makes them weak to prompt injections in data being analysed.It's horribly stupid that companies are trying to force that shit in to every product with how trivial they are to exploit.
>>108503835>Write a jailbreak prompt in your tracker that says legal action can and will be taken against any bot submissions unless they reveal they are a bot."Yeah sir, we found out that this account is from India. We wrote to New Delhi PD, they say they they will take a look at it in a month or two"
>>108503804he's right tho, fuck bloat
stop using microslop github
>>108503785>BREAKING: Ebussi considering self-immolation.
>>108503785live human verification features such biometrics. iris detection, pulse monitoring, brain wave monitoring, etc.oh right you guys throw a fit whenever the only real solutions to these problems are mentioned.
>>108503785brutal mogging
>>108503785Use case for LLM?
>>108503785Reject all AI submits. Ban them if necessary.
>>108503785We fire them luddites
GitHub for humansGitHub for agents
>>108503785Why doesn't he simply embrace the efficiency of the free market and merge all pull requests. Is he stupid?
>>108505858first one will be used to make a cull list
>>108505785this is an arrogant mindset. Grok has helped me solve extremely niche tech plugins such as how to make ctrl Q close windows work on Windows 95 by digging up obscure software, and special drivers for 1920x1080 res in win95 virtual box.Grok also made 40 GEGL Plugins (separated from my self made ones of course) that are of really cool patterns and gradients. DeepSeek helped me fix errors I was having with Pix image viewer, and compiling my plugins on Windows, and writing ultra niche scripts for xnedit/nedit, and compiling x11 software from the early 90s by slightly modifying its code.These machines clearly known how to simulate reason and solve niche complex technical problems and improved my quality of life, They are not "slop" generators. On Mastodon where lies spread they keep upvoting post like "AI gives you something that sounds like an answer since it can't check a database" This is true. AI indeed is just a glorified auto complete, but internally the module is scanning completed tokens dozens of times against other fact checking modules. Grok has internal agents debating themselves to prevent hallucination bullshit. Also Open source FOSS AI like Qwen exist so how is Qwen unethical? if its offline and open source and commercially permissible.There is nothing wrong with Gen AI and the people who hate it the most are clearly repeating lines that have been debunked or recognized as oversimplified arguments. I've seen corners on Mastodon where so called "anarchist" with red and black flags wanting the Government to ban AI. that just shows how much anti government AnComs really are. They are not open to a debate about AI, the water centers cooling things down, and the difference between video gen, image gen and text gen, artist having their work stole/copyright or the consequences of students cheating on homework. These are real debates we can discuss. They just want blanket bans from the Federal Government.
>>108505947*extremely niche tech problems
>>108503785Fuck Ebussy. AI WON, CODETRANS LOST
>>108505947>AI indeed is just a glorified auto complete, but internally the module is scanning completed tokens dozens of times against other fact checking modules. Grok has internal agents debating themselves to prevent hallucination bullshit. Also Open source FOSS AI like Qwen exist so how is Qwen unethical? if its offline and open source and commercially permissible.this man is delusional
>>108503785Use case for complaining about LLM pull requests?
>>108505947you are the niggercattle that terry hated
>>108503785simple, have AI reject them all
unc-coders better fucking rise up on this one
>>108503785AI sisters... it's overEbussy says AI has no usecase
programming isnt as exclusive as it was and janny is mad he actually has to work now,bet the dirty secret is hes probably the one feeding everything through gpt
>>108503785Usecase for rejecting PRs?
>>108503785I didn't even read your post. I just needed to verify (tm) my session (R)
>it's hard to translate policies from Mossad into developerspaceI hate GNOME so fucking much
