My company has started to ask teams to look for use cases in our day-to-day work that can be automated with AI, and I'm nervous. I heard plenty of stories from other industries about people automating their work and then be fired later.Managers are threatening us with bad performance reviews if we don't find anything and refuse to use AI. At the same time, even if we find something meaningless that looks good on paper, I'm pretty sure they'll use it as an excuse for another round of layoffs. What makes it more dreadful is that we're not getting any new projects that are *not* AI related. It's like they want to fire us by our own hands and then blame AI in the reports.How to sabotage this?
>>107073354Automate you being in meetings.> But muh meetingerinos
>>107073371I don't think it's a good idea.
>>107073354>How to sabotage this?Wait. (That may ofc mean losing your job in the interim.) Once this shit comes crashing down around investor's heads, they will be looking for someone to scalp. Pansy-ass middle manager types are the perfect mark -- and rightly so since they toadied right up to their bosses on this.
>>107073530>middle managersJokes on you they're getting rid of middle managers right now. Not all of them are fired of course but their ranks are being culled.
>>107073354Help automate managerial tasks with AI
>>107073354what do you do that they think they can automate?as a sysadmin, we're always trying to automate what we can, but there is no AI replacement comingAI does help with research lately, quick and direct answers instead of searching blogs (rip tech bloggers)
>>107073723i bet alot of those LLMs are trained on those tech blogs :/
>>107073723>they think they can automateI suppose they're after automating the whole software development process. If they knew what can be automated we wouldn't be having our jobs, but like I said, I believe they're going after numbers in reports to justify layoffs.
>>107073354give it database access and let it buttfuck your db
>>107073530I wonder how long they can keep the charade going, If it keeps up like this nobody is going to have any idea how anything works anymore, retards keep blindly trusting whatever the AI lies about. Microsofts code quality has clearly gone to shit already.Whenever I get an email from someone at work that starts with "I asked chatGPT/Copilot and it said..." I want to die, it's NEVER even close to being correct or helpful.
Make it generate a lot of unnecessary code constantly so the code base becomes incomprehensible
>>107073857That sounds like a good idea. But how do you justify having so much unnecessary code?
>>107073872>But how do you justify having so much unnecessary code?Performance. Make it unroll every single loop.
>>107073764i dont know much about software development, but the part where my helpdesk and sysadmin work is safe, is because non technical people do not know how to communicate their problems or requests properlytheir self assessments are 99.99% wrong but sometimes a random person will call it correctly, thats very rarei have to constantly remind helpdesk to NEVER trust what they say otherwise they can end up going down these rabbit holes that have nothing to do with the actual problem and wastes a bunch of timemost of the time people just point at the screen and kind of make guttural noises >problem THERE see i wasnt lying!i would imagine that software development has a similar issue with improper communications on features/adjustments/bugs etc
>>107073872If somebody challenges your MRs, ask the LLM to argue for you
>>107073953LMAO. You're good.
>>107073354Why not submit a report that automates your bosses job and advocates for them to be replaced by AI?After all, they're not going to do it. They don't have the skill set for it.
>>107073992Because it's stupid? You need to get a hold of an owner to make it work. Ain't happening, my company is not a 10-people startup.
>Argue that they should automate HR.>Build system that discriminates and brakes all kinds of laws>Sue company for that System>ProfitIt's that easy.
replace management with ai
>>107073872>how do you justify having so much unnecessary code?Have the LLM automate the code review process so there is no human to ask unnecessary questions like this
>>107073354coup the manager.
>>107073354>How to sabotage this?show them the performance of ai, kekthey WILL lose money if they go through with the idea.take a 20 bucks account at claude or somethingand vibecode your usual tasks after hoursif you succeed, congrats, youre an expert prompter and you should be safetheyre gonna use you to train the replacements, maybe supervise them evenif the chatbot crashes and burns, which is likely, thats my experience;then you go with the prompt to your lead or manager or wshatever, with a log of the promptwith annotations thorough enough so that tech illiterates can understand embecause c-class people are marketingroids, not tech savvy, usuallyidk thats what id do
>>107073354>>107074478ah, also probably dont input the actual prod code from your companymaybe just vibecode similar tasks instead
>>107074478>>107073354in my experience heres what makes a chatbot break:-exceeding context size. code that cannot get subdivided, and is long. the chatbot starts to forget things-side effects in functions. its cross-context relations, a chatbot will never be able to keep all the granularity. it also forgets things like global vars, or derives a single information in two different places-hallucinations. that one's easy, one actually has to be as un-pushy as possible to avoid triggering them. insisting on something that doesnt exist or wanting with insistence will prolly make the chatbot hallucinate-obscure-stuff statistics wise. stuff that isnt published on the internet often enough to become statistically significant. an implementation detail, an obscure framework. i had chud gpt hallucinate about c's intrinsics, but also shit like morrowind scripting stuff just because of the lack of accurate documentation
>>107073354Tell them HR and Upper Management. The most worthless and expensive employees in the building.
>>107073354why sabotage when you can make bank by just spinning up a competitor that makes a big deal about how you don't use ai.really this is the ideal time. ai hate is at an all-time high. someone could be making bank right now by just creating the organic produce of the computer world and society would immediately cum and all switch to it.
Anons itt giving guides on how to make ai perform like shit while gpt and gemini can't handle webshit, or even creating a docker image properly
>>107075353but the idea is to convince beyond reasonable doubtyou take the bestand then make it fail in reasonable scenariosits to remove the idea of >yeah but what if we took the premioomnonono, heres premioom, spreading its cheeks, kek.if we go that way we lose money kind of dealthe alternative is that it actually performs to a satisfactory level and then op has a (probably incomlpete) list of traps to avoid and a leg up in his ai reportits nto the good ending but its not the worst either, methinks
>>107073830>I want to dieUrmm, wouldn't it be better if THEY died, and you lived, anon?Regardless, you're correct. The incompetency crisis is real. A famous biblical adage is "Sow to the wind, reap the whirlwind." (((They've))) sown to pozz again and again in their futile attempts to destroy Whitey. They are in the initial phases of reaping what they've sown. Simple as.>tl;dr"Buckle your seatbelt, Dorothy. Cause Kansas is going bye-bye."
