I didn’t say the don’t work. They’re just mainly useless. For coding? Nice for quick prototyping, seeking alternate approaches or such. Real code? Rarely working and in the end you spend more time fixing stuff than it initially saved.
For everyday research? It’s helpful sometimes, but mostly plain wrong. That one sure isn’t inherent of LLM but the data it feeds on.
For what it’s great is translations, summarizing stuff or a nicer web search. And my own locally run for the smart home. For that it’s a blast. Like “close gate, open front door, turn on all lights at 50% in a lush shade of green, and turn on the coffee machine”. Absolutely awesome.
But paying for that? And even like 30 moneyz? No way. Don’t get me wrong, I fucking love all tech and LLMs are kinda a dream coming true. But it’s still not really there. My last session with Claude ended in the “AI” giving up and asking ME how I solved it. Only to then admit my answer would be useless as it is forgotten anyway after that chat 😁
I have no political agenda and surely don’t lack experience. I don’t even see LLM as an enemy. I wish it was actually intelligent and really helpful/timesaving more than 50% of the time.
Not questioning generative powers for video or image or sound. They’re awesome there and totally timesaving and extremely helpful.
stressballs@lemmy.zip 1 week ago
Yeah screw all those disabled people who benefit from it. We’ll just group them in with the other counterfactuals you have to ignore to maintain an opinion like yours.
Dyskolos@lemmy.zip 1 week ago
I have no idea how disabled people have anything to do with the current quality of LLMs. I didn’t say it’s not helpful at all, i didn’t say it does not work. Can you americans just think in binary dimensions? Some people don’t have a political agenda and just see things like they are. Like I do. And to me, LLMs are 90% useless, 10% incredibly helpful.