But yes. Exactly in the use of “Artificial Intelligence”.
Artificial Intelligence is a wide field, consisting of a plethora of methods. LLMs like ChatGPT are part of this wide field, as per definition how researchers are describing the field.
The “intelligence” part is an issue though if taken literal?., since we have no clear definition of what “intelligence” even is. Neither for human / natural intelligence, nor for artificial. But that’s how the field was labled. We have created a category for a bunch of methods, models and algorithms and sticked “AI” onto it. Therefore I stand by what I have said before:
It is AI.
Due to the lack of a clear definition for “intelligence” I would coarsely outline AI as: mimicking natural thinking, problem solving and decision processes without necessarily being identical. (This makes it difficult to distinguish it from plain calculators though, so a better definition is required.) So if we have a model that is able to distinguish cat pictures from non-cat pictures, that’s AI. And if we have “autocorrect on steroids” (credit to Dirk Hohndel) like ChatGPT, that matches the text comprehension skills of 15 year olds (just an example), then this too is AI.
One fundamental part of “intelligence” is being able to come up with independent thoughts. Another is to be able to think critically about those thoughts. LLMs cannot do either.
I was going to wholeheartedly endorse your comment and then you ruin it in the last sentence with
that matches the text comprehension skills of 15 year olds (just an example), then this too is AI.
It feels like you know what you are talking about, but then confuse the successful statistical analysis of text as “comprehension” which is just plain factually wrong.
mavu@discuss.tchncs.de 5 days ago
Not if you use “AI” as the abbreviation of “artificial intelligence”.
If you use AI as meaning “what chatGPT & co are” then it’s a trueism.
Zacryon@feddit.org 5 days ago
But yes. Exactly in the use of “Artificial Intelligence”.
Artificial Intelligence is a wide field, consisting of a plethora of methods. LLMs like ChatGPT are part of this wide field, as per definition how researchers are describing the field.
The “intelligence” part is an issue though if taken literal?., since we have no clear definition of what “intelligence” even is. Neither for human / natural intelligence, nor for artificial. But that’s how the field was labled. We have created a category for a bunch of methods, models and algorithms and sticked “AI” onto it. Therefore I stand by what I have said before:
It is AI.
Due to the lack of a clear definition for “intelligence” I would coarsely outline AI as: mimicking natural thinking, problem solving and decision processes without necessarily being identical. (This makes it difficult to distinguish it from plain calculators though, so a better definition is required.) So if we have a model that is able to distinguish cat pictures from non-cat pictures, that’s AI. And if we have “autocorrect on steroids” (credit to Dirk Hohndel) like ChatGPT, that matches the text comprehension skills of 15 year olds (just an example), then this too is AI.
princessnorah@lemmy.blahaj.zone 5 days ago
One fundamental part of “intelligence” is being able to come up with independent thoughts. Another is to be able to think critically about those thoughts. LLMs cannot do either.
mavu@discuss.tchncs.de 4 days ago
I was going to wholeheartedly endorse your comment and then you ruin it in the last sentence with
It feels like you know what you are talking about, but then confuse the successful statistical analysis of text as “comprehension” which is just plain factually wrong.
sad. so close though.
Electricd@lemmybefree.net 5 days ago
Enjoy being downvoted for being right
Zacryon@feddit.org 5 days ago
I don’t care about votes. I just hope that people start to comprehend this field a tiny bit better .