News Researchers hope to quash AI hallucination bugs that stem from words with more than one meaning

[Admin]

Semantic entropy is when we use words that have multiple meanings, thus leading to confusion. The researchers want to use this as a way to estimate the chance that an AI is hallucinating.

Researchers hope to quash AI hallucination bugs that stem from words with more than one meaning : Read more

 

[Ralston18]

What I red was that AI cannot understand via context.

If that is reel then it could be difficult to determine if weather or knot some AI response is an hallucination or knot.

Four example: how do I fix the breaks on my car?

Answers could go right down into the rabbit whole.

Such errors could be two much or to funny for reeders to bare.

 

[USAFRet]

Four example: how do I fix the breaks on my car?

Wot do u mean? Going threw my garage, I c sum stuff on the ground. Can I fix it?

 

[Findecanor]

Oh. I think someone could have found a use for
Lojban, — a synthetic language made especially not to have any syntactic ambiguity.

 

[Ralston18]

Eye u can! Get a big mall.

 

[USAFRet]

Eye u can! Get a big mall.

wot do u mean? y u bean so mean to me? I was just askin a ?

 

[bluvg]

Oh. I think someone could have found a use for
Lojban, — a synthetic language made especially not to have any syntactic ambiguity.

English is awful in so many ways. How many mistakes (big and small) could have been prevented by addressing this flawed subsystem we rely on heavily?