Comment Are you a troll or just clueless? (Score 1) 51
Can you spot a trend? Even if your experience shows AI lacking now let's look at where it was at 3 years ago: zero. So in three years it went from 'zero' to 'useful sometimes/as good as recent graduate' and this is the worst it will ever be. If that trend continues at pace you are out of a job in three years. Things slowing down you say? The breakthroughs they keep a coming in software, like increased memory efficiency: https://techcrunch.com/2026/03... and hardware like Rubin rack computing units: https://www.tomshardware.com/t... Now here is the grand problem: you need to spot your jobs are going away to change how you vote BEFORE your jobs actually go away, and if you can't do that the average joe surely can't either
Everything you said is true about theoretical AI, but not about real-world AI...or at the very least Claude Opus/Sonnet 4.6. It hasn't improved in any measure that I can identify in the last 2 years. I've been using it every workday, by mandate from my employer. So I am not a naysayer...I'm a frustrated user.
Look, you can show me bullshit theoretical papers and press releases all you want. There's a simple test that will prove either of us right or wrong and that's the real world and the market.
ChatGPT was released nearly 4 years ago. By year 4 iPhones were largely successful and disrupted many existing industries. My favorite example was the gay bar scene was annihilated by grindr. Why overpay to go to a windowless bar, face potential discrimination, and deal with sexual harassment and frustration to date a limited pool...when you can just check for available dick on your iPhone? It was starting to disrupt cabs. It annihilated sales of GPS units and even cameras. And iPhones were EXPENSIVE and had a fraction of the penetration ChatGPT does. FAR more people have used ChatGPT than have even used iPhones today, let alone 4 years into their existence....yet few come back.
LLM-based AIs? The only people making money in this gold rush are the pick and shovel vendors. No one is launching new LLM services that people actually want (everyone hates chatbots). I can think of a dozen KILLER usecases for easy money if AI was as good as you said. For starters, how about resurrecting ancient video games? Epic could spend 20k of Claude's time and easily reboot a major video game, like Unreal Tournament in UE5....if the shit actually worked. EVERY compiler maker could optimize binaries further if LLM-AIs actually worked. While no one pays for compilers these days, every hosting company would love to make their binaries 20% more energy efficient...get more hosts on a machine, get better response time, etc. Or more obvious use cases are apps that convert legacy COBOL code or any other language to the modern language of your choice (rust/Java/C).
Put simply, LLM-based AI resembles bitcoin more than it resembles the iPhone or internet or even minor tech revolutions like cloud computing and big data. Right now, the only people profiting off it's introduction are the pick and shovel vendors...the AWSs, the nVidias, the AI providers. Once I see pioneering businesses doing useful things with AI, beyond selling AI, then I'll start believing the hype.