What % of AI optimists are grifters?What about those will AI psychosis?Those with not direct monetary stake and a functioning brain?I am curious what you all think because it seems like most people hyping AI who aren't directly invested or working on an AI startup seem to be a bit delusional at what the bots can do and what they've accomplished with the bots.
Holy fuck I made many typos.Never phoneposting again (I will, though)
All of them. Dude. All. Of. Them. How many pump and dumps is it going to take. Grifting is the only business model of 99% of startups. Aka small businesses. Even the fairly benign ones just want to do a few years of lunches on the bank's dime without ever contributing anything to society.
>>108185289On this site at least they're all trolls. That's why every post is "haha get fucked (job associated with the board), AI is going to replace you"
>>108185289Everyone who isn't building is a grifter. There I said it. Everyone knows its true.
Luddite cope thread.AI GODS WON
One day the technology will get there and everyone will forget and move on like nothing happened. /ic/ /lit/ and /mu/ have already been automated and they barely even talk about AI anymore, meanwhile usable AI 3D models are over a decade down the line and /3/ never shuts up about how they're going to be replaced.
>>108185289a few months ago i tried out having deepseek chat help me with coding. it was very hit or miss, with hallucinations being a semi common problem. i realized after working with it for a while though that 90% of the time it sent me down some wild goose chase, the problem was that i had left out some detail in another file that i didn't know was related, so arguably those issues were partially my fault. but on the other hand, if i knew the related stuff was the issue, i wouldn't have needed ai so it's kind of a catch-22.it was very helpful when just using it as a google replacement or if i gave it a chunk of code and asked it how it worked. but for actually writing code i wasn't in love with it.a few weeks ago i finally ponied up 20 bucks to see if the claude code shit everyone was hyping up was actually any better. it seems way more useful because that "not having enough context" issue i had with deepseek was not longer a factor since it could just read everything in the repo to see what it needed to do. while i did audit all of its proposed changes, it pretty much never fucked up. there were times where what it gave me wasn't feature complete and i'd need a few more prompts to get what i was looking for, but i've yet to have any issues with hallucination. i assume this is less about about claude being super special or even the agentic shit, and more just that it had full project context. that said, these was a relatively small project with probably 5k lines of code. i assume if you've got some huge project with a million lines of code, having full context isn't possible in the same way and these benefits fall off.at this point when somebody tells me AI is dogshit and hallucinates garbage constantly, i just assume its someone who has only played with free-tier webchat and not the newer toys. not that i fully buy into the hype, but i at least find it somewhat plausible now when some of these companies say AI writes like 70% of their code
>>108185978oh but despite finding claude code more useful than a lot of "ai is shit" people, i still absolutely do not trust a word any ai ceo says. they've already spewed so much bullshit about AGI being "right around the corner" for years with only incremental improvements. altman and co are hypemen first and are completely desperate to keep their EXTREMELY unprofitable companies rolling while they burn through billions of dollars. i'm just saying that some of the tools they make aren't complete crap like some ai doomers claim
It doesn't matter how anyone "feels" about it. It happened, is happening and will happen and it will only continue to happen more and more.Cat's out of the bag.
>>108186064>>108186066sure>deep seek chati was setting up some docker compose files for a web app and it helped with basic setup but hallucinated some parameters that i figured out didn't exist after wasting a lot of time.later on i was trying to configure the web server and it pushed me to set up static site generation. a lot of my pages were dynamic though (needed to pull fresh data from database before render) so this ended up being a huge waste of time because it didn't know about those dynamic pages since it hadn't looked at them that session>claude codei had another web app that i was considering swapping from react to svelte. not a huge app, just six pages but each had some unique functionality to them. i decided to see what claude code could do so i just prompted it to port the frontend from svelte to next.js and it did it in one shot with no issue. some of the components it made were pretty long, so i ended up having it do a small refactor to break them down, but there was nothing technically wrong with the code.later i asked it to make a basic multiplayer battleship clone with web sockets and it pretty much one shot that as well. it wasn't completely perfect, like i had to prompt it further to handle stuff like "if player 2 disconnects, tell player 1 and let them go back to the menu", but the core functionality was there. and this was with me playing pretty fast and loose with the prompts and basically just giving it high level requirements, if you used it more intentionally like planning the code out yourself and prompting it for individual functions i'm sure that gives you a lot better control. i just wanted to see it from a true "vibe coding" perspective
>>108185978opencode
>>108185289Has to be a lot of grifters and Indians. I love the tech, toy around with local models all the time, and would love to work in the industry. But god damnit if the current course of this sloppy shit doesn't make me pissed off to no end.
>>108185289AI as it's being used right now is nothing but a corporate pyramid scheme trying to convince people that chatbots can replace the labor of humans.What we need to be working towards is making AI that can pilot a spaceship to explore and navigate outer space independently, because God knows we're nowhere close to being able to do something like that in person.
>>108186314i think it kind of depends. like in my example above, i'm not really pushing software forward at all, i'm using existing tools/frameworks to make stuff that has basically been done before. i think AI is obviously going to be very good at that since it knows the tools and design patterns really well. not that its training data has an exact copy of what i needed and just had to regurgitate that, but it had all of the right pieces and knew how to put it together. i don't know how useful AI is for doing actual cutting edge shit that there isn't a well defined playbook for.the thing is, like 99% of enterprise software development isn't doing anything cutting edge, so i don't think that limitation is an issue for a lot of companies. i don't really think we're heading to a future where SWE is purely a hobby and not a job in the near term, but some companies are definitely dipping their toes into that water. but then you have other companies that run their stuff on the same shitty infrastructure for the past 20 years and barely think of AI beyond seeing it talked about on super bowl commercials, so even if the tech was perfected today they'd be slow to adopt
>>108186424The sad thing is that we already can make very useful models that can automate shit, or just act as a simple knowledge base for certain info if we just finetune them on the right information. But few seem to be trying to gear a company towards that purpose, and companies aren't buying that because it's boring and not flashy.Big massive slop models that try to emulate a person is what gets attention and investor money.
>>108186424>>108186475Yeah, right now rather than doing shit like trying to solve a cure for incurable shit or trying to crack a physics code that impedes space exploration or any other human issue investors and companies are just using them on how they can automate jobs and fire their workers to make line go up, that's not happening.
>>108186314>>108186472i'd mention as well that AI anxiety and tech job market anxiety is kind of getting conflated these days as having more of a cause and effect relationship than they might really have. not saying that the AI hype craze hasn't been bad for the job market, but the market really went to shit in 2022 because of interest rates, not AI. federal fund interest rates were extremely low from around 2008 (in response to the housing market crash) to around 2022. low rates means venture capital can get less risky loans, which means they can be irresponsible with spending. that meant tech hiring was insane in the 2010s because VCs were happy to throw money at any stupid idea. but during covid, the fed was irresponsible with how much money it printed and that caused inflation, so around 2022 they raised rates a lot to compensate. this meant the free money train for VCs was over and mass layoff in tech ensued. it's basically a coincidence that this happened around the same time openai release gpt 3.5 and ai became a hot topicnow interest rates are still pretty high and the tech job market is still in the shitter but i feel like people have formed this narrative the tech job market is in the shitter because we've replaced everyone with AI. again, not saying AI hasn't replaced some people or that the tech market would magically bounce back if interest rates dropped tomorrow, just saying the current job market and rise of AI aren't necessarily as tightly coupled as people seem to assume
>>108186533Don't forget that covid also ended and everyone wasn't sitting inside playing on the internet anymore, so there's less money to go around now in web development, which is almost all programming these days.
>>108185289>What % of NFT optimists were grifters?about the same.
let's say you do 6 things in a loop on your computer:>play videos (voice mostly)>listen to music (soft background)>surf websites (text, summarization)>chat with your friends (text, voice)>play games (full screen attention)>write code in a development environment (4+ windows tabbing between them)instead of fiddling with your mouse or keyboard you can just use one or two shortcuts to bring up these different situations on your desktop and speed up the process. you read 5-10 news websites a day? AI can read it for you and summarize everything into a few paragraphs and provide you full context if you want to read more. You're writing code? AI can write thousands of lines a day, build test cases, run the entire thing and give you a summary report of any progress it made towards your goal - you can drop it at any time and take over or inspect what is going on, but mostly your role is as supervisor or manager. Have 30 videos in your backlog across many tabs in your browser? AI can combine similar videos, do video to video summarization, video to text summarization, and instead of having to watch 15 hours of footage at 3x speed to keep up you might watch 15 mins of an overall text and video summary and then you can be more selective with full length content. Like a particular artist? Have the AI generate a unique copy of their sound, mash it together with another band you like, now you have unique music generate on the fly just for you. i think the endgame is to replace all internet with generatenet, and move the web backbone to gpu datacenters where you can 'store' the information cheaply and retreival is just replaced to cloud and local generation of content from storage prompts.
>>108186629That sounds like a dogshit way to consume/create media or art and work, no thanks.
>>108186656the alternative is google goes bankrupt hosting everyone's vlog for free on youtube, goes to a paid uploader model and collapses cratering the entire industry. AI might just be a smokescreen for everyone to raise money to build data centers to host everyone's social media footage. i tried to give a plausible future scenario where AI is integrated into daily computer use overtly for most desktop users.for smartphone users it might just be they consume AI content non-stop and nothing changes for them. AI gfs/dating, AI tiktoks, AI paying them to watch ads and like videos in exchange for free burgers.>>108186665can always opt out, like the people that refuse to buy digital games or subscribe to netflix. most people follow the path of least resistance, is uber better than a taxi? are airbnbs better than hotels? is postmates better than a pizza guy? we might have a trillion dollar IPO with OpenAI later this year, the FOMO is real just look at the market cap of the top 5 companies.
Where's the people that see technology as what it is (tools for creation and optimization) without giving it some personal value?
100% are grifters.AI used to be fun project for students and research topic. For the past 5 years it's nothing but a grift. AI is seen as magic by normies and it's easy to fool them into thinking AI is anything but what it actually is.
>>108186690You're being incredibly reductionist and jumping to the doomer conclusion, even if you do want this dystopian-like AI generated controlled future to happen (and you're genuinely insane if you do) not everyone will be on board with it,Public transportation like city busses and trains/subways still exist and has since and will for decades, staying at someones house or renting a room at a cheap motel will never ever probably go away until we find a cheap and infinite housing solution (never probably), and delivery/courier jobs well let me ask you, are you willing to be a delivery boy when AI eventually "supposedly" takes your white collar (which most of /g/ is scared of losing atm) job? Because if you say "no" well I got some funny news for you.
Any productivity gain that doesn't fail to materialize will be absorbed by the humans getting lazier.Much like the desktop experience hasn't improved for the average user in the last 15/20 years. Even though hardware has become incredibly faster, the gains are absorbed by mediocre software.
>>108186071Nothing has happened, nothing is happening, and nothing is going to happen. You are delusional and suffering from AI psychosis. Seek professional help NOW!
>>108185289What % of anti AI people do it just because they have vested interest in AI not becoming capable of what they do?
>>108185289Well, the ultimate end of the current AI push is the mass forced retirement of the executive class, the final destruction of the urban office real estate market and thus the destruction of cities like NYC as a whole. So you can be optimistic about AI for numerous different reasons. Almost all of them are malevolent and take advantage of the naivety of a class you're looking to make your victims, though. The funny thing is that the executives think they're going to cull the labor force beneath them when AI is actively aimed right at them. Big red dot on their foreheads and they think it just means they're cultured.
>>108189272https://youtu.be/zGZqnc6_5To
>>108187232Unclear.I am an AI pessimist by most people's standards, but I don't think its entirely useless.I don't think it will eliminate SWE jobs by huge amounts or 10x productivity, but it's decent for some use cases, especially search engine.I work on the non-technical side of tech companies.
>>108185289there are:- the jeet who thinks using ai will make them mimic humans better and so get a better chance at immigration or an offshoring job- the neets who believe that finally all will be unemployed losers like them- the middle management that is delusional about being able to finally get rid of programmers- the retards who barely learned anything and coasted during their education praying for a chance to compete with ones who acutally did the work- grifters who are 12 months away from replacing white collar jobs (for the last three years)no point in trying to guess percentages honestly, just bring up that inference alone operates at a loss or that there are no real plans for profitability and the optimist dies instantly (leaves to post a new retarded "replace when" post)
>oh no, smoke and mirrors made to fool investors instead of actual innovationWe've been in this pit for like 20 years
>>108185289>>108185294>>108185311>There are "AI Experts" who always predict AI will never be able to do anything better by next year>They get BTFO'd. The ai increases its skill level significantlyand then every retard here goes "ITS LE BUBLE!" Fuck yourself midwits
>>108185294medsNOW
>>108192562This. Just look at how far it came in just 11 months. We're looking at AI generated movies before 2025 and replacing 90% of the white collar workforce before 2026.
>>108192562Not worth the investment cost
>>108192717Yea its a buble like when bitcoin went from 1 cent to $100. That 100 dolla bubble
>>108192562The CEOs told me in 2023 all or most coders would be replaced by 2025.Just two more weeks and two trillion more dollars, for real this time.
>>108192799>he hasn't been replaced by barbara and her $5 chatgpt pro subscription yetbro... just use claude save yourself
Recognizing that AI isn't a bubble that will pop and disappear makes you an "AI optimist"? People in the early 70s said computers were never going to be in every household.People in the 1990s said the internet was a fad and that no one would trust buying shit online. Now you have retards saying AI is useless and once the bubble pops it will go away.
>>108192783What's it at now?
>>108193024Mind you it was $119K in December so this is technically after the "bubble" burst.,
>>108193033Six sevenn
>>108185978>at this point when somebody tells me AI is dogshit and hallucinates garbage constantly, i just assume its someone who has only played with free-tier webchat and not the newer toys.I think it's amusing to watch people essentially self-sabotage by denying reality and spreading that denial. AI is one of those things where I don't even feel like I should try to convince people that it's going to be useful, because it's going to happen. It's inevitable. What AI can already do is going to be a multiplier to human labor in a whole bunch of fields. The only uncertainty is about whether it's going to get even better than it already is.
>>108185289I was a pessimist for a long time, but the tools have got good enough now (for coding but also for basically anything that can be done in a terminal) that I can no longer reasonably deny that everything is changing. I don't know if it'll lead to AGI, but it no longer feels impossible and even if it doesn't happen, the ANI we have now is disruptive enough.To anyone who still hasn't really learned these tools beyond occasionally using chatgpt, now is the time to git gud. Get something with good tool use support. Learn about context management.Understand that most LLMs are autoregressive and what this implies (e.g. if you want citations, the sources need to be included first so they can be used to write the response. Generating a response and then asking for sources afterwards gets you hallucinated garbage.)
like 90%, the rest are tech illiterate retards that have no idea how the technology worksThere is no reason to be excited or hype up a stochastic parrot that hallucinates simple questions and contradicts itself in order to kiss your ass
>>108185289It's the usual 80:20 rule.20% of grifters. But they are responsible for 80% of the (fake) engagement.
>>108185978Pretty accurate.
low intelligence: linear growthpeak midwit: exponential growthactually smart: logistic growth
