Is AI really as big as most people claim?
>>62134920Yes. What is less certain is how big it will be in a decade. Or next year.
>>62134920I am an "elite" dev - I know all the best practices, how to make code re-usable, how to make it easy to maintain, fast, secure, blah blah. I haven't written a single line of code in the last 3-6 months and Opus 4.7 is better than 4.6. I do still need to know what to ask it and sometimes, rarely, catch critical bugs that it says don't exist. You still need people to build high-level (2-5% of jobs) pipelines ( CI/CD, Terraform, real architecture which is like 3 people at a huge bank). I feel like it's overstated because it's made to replace devs, it's not general. What it will do is kill software development, UI/UX designers (already did), and some adjacent industries.To me this is very important, because for a while IT was a way to get out of poverty without connections. No there is nothing that I know about. Maybe nursing, but nursing is literally wiping shit.As far as investing in it, you will still need to make the right picks. There will be a lot of shit "AI projects".
>>62134969It seems Opus got significantly worse in the last 1-2 months, can you confirm this?
2 more weeks until le bubble pops!!!!!!
only "coding" I do is making autohotkey scripts and damn ai still needs lots of prompting and correction . I can't imagine what it's like for something more advanced
>>62134920No, people are investing trillions of dollars in it just for fun
>>62134920NVDA has a lower PE than Walmart. There is no AI bubble, we live in AI reality now. Which means NO JOB 4 U.
>>62135241No. It got better in my experience.>>62135274I don't believe you for a second. Give me an example, I will literally have it write the script for you for fun because it's one prompt and requires zero work.>>62135248I do think all the bullshit AI companies will pop, AI is literally coding / website creation tool. It's much worse for everything else. This is because it has a huge training set for websites/apps.
>>62134969maybe i'm ignorant but AI seems to be doing technologically what Adderall does on a biological role. It's increasing people and corporations output capabilities. but there seems to be an assumption that increased output = more productive by default. At least in my job we're using AI as a crutch to overcome issues that could have been avoided in the first place. Things like using agents to automate bloated processes instead of just streamlining and cutting down the middlemen. It's like using uber and robot cars to solve the lack of public transportation. Like yeah it's cool but also stupid
>>62135274AI is fine for simple stuff but long-term complicated projects can't be vibe-coded and still needs a competent dev to figure outbetter off just using it as an assistant instead of everything
>>62135740Right, you can have Indians now working 20 minutes a day and wasting even more of your time with "well, it's best practice, here is a quote from GPT, so it must be right". You still need people to be actively engaged to get value out of it. It makes the job even worse, because now you basically have 0 skill, and are a retard(AI)-wrangler.>>62135756It's a little bit of both, the key point here is that the dev is no longer figuring out dev shit. They are figuring out how they are going to host, deploy, what framework to use, etc and doing SOME code review. Basically they are architects and AI are the mids. Unironically, they are also QA now.It was also true that product owners / other "managers" weren't just "not-coders", they also just say high-level hand-wavy shit which doesn't cut it with AI either.The problem is that you now have NO way to prove to an employer that they need you except in rare cases where you have technical high-level directors/VPs, which does happen actually.
>>62134920its more useful than most people.i mean seriously ask a person questions then ask the LLM. the LLM will generally be better.
>>62135756its def changed SWE jobs forever. the uninformed point to vibe coding but that misses the mark. SWE is changed.
Whenever you hear someone deep into the AI rabbithole speak about it it becomes obvious how desperate they are for it to be the future and for them to have been part of it early. The exact same talking points and logic used for NFTs and VR only a few years ago. Pathetic drivel of drones devoid of humanity that are so stuck in the future they imagine that they refuse to see the present.The difference from VR and NFTs is of course that AI actually is useful, just not nearly as much as these people believe. AI isnt a technological innovation at the same level as the internet or computers or electricity. Its more at the level of photoshop or the microwave.
>>62135796that's a people issue
>>62134920Yes the folder I keep my gay furry porn gens and checkpoints is almost 200GB, it's quite big.
>>62134969>>62135546>>62135784> elite dev
>>62135796Only if you surrounded yourself with retards. If you're surrounded by actual intelligent people you'd see that modern AI is still very much dogshit.>t. Surrounded by PhDs in a research capacity. AI is okay for entry level mindless drones and that's about it.
>>62134920>Is AI really as big as most people claim?No. Do some due diligence and you'll see that AI is a colossal failure. E.g. https://github.com/sgl-project/sglang/pulls https://github.com/vllm-project/vllm/pulls - most popular AI inference engines have 5 thousand unresolved pull requests combined. If AI was as hot as they claim, would most fundamental AI tools have 5 thousand unresolved pull requests? No, AI would shovel through them in a day. But it doesn't because it can't be trusted. The bottleneck moved from writing code to verifying it works and merging it, and writing code outside prototyping was arguably never a bottleneck to begin with. All you're getting is more trash. Instead of writing 3 page documents nobody reads, people are now shitting out 15 page documents nobody reads. You're getting piles and piles of software that does the same thing that can be done with the existing software. Hundreds of agentic harnesses that all do the same thing - HTTP requests and JSON parsing. Tons and tons of new websites nobody cares about, that you could have slapped together with Microsoft FrontPage 2003. It's all a big circlejerk of wasted effort that results in nothing of real value, and nothing that you couldn't have done before. It lowers the bar so much a random idiot can now shit out a website, all you get is more trash for Google to inder. It allows senior developers to leverage their knowledge and never write code manually again, all you get is an already half-dead industry that exhausted all that it can do, exposing itself as a complete farce.>t. senior C++ dev who stopped writing code manually 8 months ago
>>62134920No, it's worse. It will decrease wages and catastrophic tech failures, directly and indirectly.
