Is Ai right all the time?

Henrywrites

Henrywrites

Well-Known Member
Full GL Member
Credits
2,167
Artificial intelligence has broken down the ability to research in simpler ways. Do you think Ai can be seen as very good and can provide the right information, or does it make mistakes sometimes?
 
I think AI can make mistakes sometimes if it gets the answer from a less reliable source like social networks which can have a lot of fake answers. AI can also provide good answers to text questions if it gets the answer from a more reliable source like an Encyclopedia or more trustworthy news website.

I think AI sometimes have problems creating more realistic computer graphics because AI is not as good at drawing human fingers according to tech blog posts I read about AI drawing human fingers.
 
As someone who uses LLMs daily, they are wrong all the time. There's even a term for it. Hallucinating. Doesn't mean they're not useful. It just needs human oversight before being used.
 
Burnsy said:
they are wrong all the time.
Click to expand...
They are not wrong all the time, that's unhelpful.

AI is a great tool but its just a tool and it requires a skilled wielder to use that tool effectively. This is why there is a disclaimer at the bottom of ChatGPT saying, "ChatGPT can make mistakes. Check important info."

Always verify the output, especially if its complex or an 'opinion'.
 
AI can be right if given a verifiable fact. A very good example is coding help. It needs the right inputs to get the right outputs, but it can be right. It is mostly wrong when the input has some errors to it.
 
Absolutely not. I cought him making mistakes plenty of time. Let's say he gives me code that works half of it, then I tweak him and tell him what to edit, then he fixed that but the code I told him to tweak, he makes the same mistake again..
 
We can't expect AI to diagnose issues using a human mind since that's not what it does. It doesn't have the ability to diagnose complex issues, just to pull answers from search engines and interpret the information to fit the question.
 
_Al said:
We can't expect AI to diagnose issues using a human mind since that's not what it does. It doesn't have the ability to diagnose complex issues, just to pull answers from search engines and interpret the information to fit the question.
Click to expand...
Yes, but the thing is he is using what other people said on different websites and accepts things to be truth even if they are not. I saw this plenty of times when I asked ChatGPT. He doesn't have any critical thinking because he doesn't think. He just does a research and then put any info he found on the particular subject.
 
froggyboy604 said:
I think AI can make mistakes sometimes if it gets the answer from a less reliable source like social networks which can have a lot of fake answers. AI can also provide good answers to text questions if it gets the answer from a more reliable source like an Encyclopedia or more trustworthy news website.

I think AI sometimes have problems creating more realistic computer graphics because AI is not as good at drawing human fingers according to tech blog posts I read about AI drawing human fingers.
Click to expand...
I agree with you. AI is not perfect for now, and there can be mistakes. I've not used it for something that complex, but with "normal" questions I've never had problems.
But yes, with pictures, it's funny. I always have something wrong with the hands of generated people
 
Nope. It can make mistakes.
However it's slowly but clearly evolving. May be in the future, I wouldn't be able to say the same but for now, it can make mistakes
 
obviously not.. not all the time, but I think there can be some expectations made when its not harmful. like if its something a human can do, like labor or some shit, then ai shouldn't be needed. smaller things like door dash and anything that falls under the umbrella of "things humans are to lazy to do but easily could" then sure, no problem. :,3
 
You must log in or register to reply here.
Back
Top