Name one category of tasks that you would feel confident it can perform with at least a 90% success rate.
Comment on Anon is waiting for Japan
uranibaba@lemmy.world 4 weeks agoA lot of people here on Lemmy keeps saying that AI is bad because it failed one task it wasn’t built for. Or because it can’t do everything. I don’t get it.
taladar@sh.itjust.works 4 weeks ago
WorldsDumbestMan@lemmy.today 4 weeks ago
Making up bullshit. It never failed me yet…
datendefekt@feddit.org 4 weeks ago
Translating text. I recently had to translate a bunch of architectural requirements into english. I honestly don’t know if deepl.com uses GenAI or what. But the job was pretty much copy and paste. Occasionally I had to change a word because the connotation was a bit off, and a few times it got confused with tangled, run-in input and I had to rephrase whole sentences. I’m a native speaker in both languages and I estimate it would have taken me at least three times as long.
uranibaba@lemmy.world 4 weeks ago
Improve text that I have written. With improve I mean change the text accodring to the specifications in the prompt provided to the AI.
Here are a few more:
- Since ChatGPT has parsed most of the puplic internet, it can be useful when trying to find information about something that is very obscure - (for example settings for an old or not as used software), where my ordinary searches has failed me.
- In addition to the previous one, it can be good way to find better search terms.
- Write repeated text with slight variations that I could do myself but an AI can do instantly.
- Translate and XSD (XML specification) into another structure (for example classes when writing code).
- Create macros for World of Warcraft.
- Explain errors outputed buy some software (ties into the first two).
I am sure there are other usecases that I could not think of.
Is the money, time and energy spent to create a tool that can do this worth it? That is perhaps the question want to ask and perhaps your answer is no.
lightsblinken@lemmy.world 4 weeks ago
maybe go for a “return on investment” angle… for the amount of literal billions we have thrown at it, perhaps its ok to expect more. if you gave me a mere couple of billion, i’d make healthy lunches for school kids to foster education and health outcomes (2-4-1!)
MouldyCat@feddit.uk 4 weeks ago
How many billions (in today’s money) were spent on going to the moon? What about the billions poured into refining the internal combustion engine? The billions that have gone into making and running massive particle accelerators?
Technology is constantly advancing and we often don’t know where it’ll take us until we get there.
uranibaba@lemmy.world 4 weeks ago
I haven’t spent billions on it, so it is not a bad ROI for me. Perhaps it is for those who has invested in creating OpenAI, LLama etc, I am not one of them.
Spending the same amount of money to create a better world would be ideal. But if the money was not spent on AI development does not mean that it would be spent on anything better. That is also a discussion about the money spent on AI and if it has been worth it (a discussion very much wroth having), it does not diminish the usefulness of AIs.