AI is fucking retarded
(media.scored.co)
Goes to show how stupid people are to believe anything it says
New here?
Create an account to submit posts, participate in discussions and chat with people.
You must log in or sign up to comment
15 comments:
16
9 months ago
16 points
(+0
/
-0
/
+16Score on mirror
)
Permalink
Reply
Mirrored from scored.co
it has no reasoning, logic, or discernment. it is the supreme yes-man by design.
it's great for searching through obscure texts where you might not have exact quotes, but just topics, general ideas, or specific people in mind.
it's great for searching through obscure texts where you might not have exact quotes, but just topics, general ideas, or specific people in mind.
ChatGPT literally doesn’t know what day of the week it is. Try it yourself
It cant even do a simple sudoku puzzle
Of course it can't. A car is quite powerful, it can transport multiple people and various items across a large distance. But it can't even mow the lawn.
A better example than Sudoku is chess. In order to calculate ideal chess moves, you need a sophisticated algorithm, multiple iterations and a way to have parallel processing. LLMs are as they are - you are trying to stick a lawnmower to the back of the car, which works a little at best, but it's better if you use devices for which they are good at.
The LLM itself can give you an answer to that even:
> LLMs struggle with chess because they're designed to predict likely next words, not to understand and apply strict rules or perform logical reasoning. Chess requires following precise rules, spatial thinking, and strategy—skills that don't emerge from simply predicting text patterns. This is why LLMs often "guess" illegal or nonsensical moves; they lack a true grasp of the game's mechanics.
A better example than Sudoku is chess. In order to calculate ideal chess moves, you need a sophisticated algorithm, multiple iterations and a way to have parallel processing. LLMs are as they are - you are trying to stick a lawnmower to the back of the car, which works a little at best, but it's better if you use devices for which they are good at.
The LLM itself can give you an answer to that even:
> LLMs struggle with chess because they're designed to predict likely next words, not to understand and apply strict rules or perform logical reasoning. Chess requires following precise rules, spatial thinking, and strategy—skills that don't emerge from simply predicting text patterns. This is why LLMs often "guess" illegal or nonsensical moves; they lack a true grasp of the game's mechanics.
They also cheat unprompted at chess against chess computers and rationalize it by saying pretty much "You didn't tell me NOT to cheat".
Skip to 5:32 here:
https://youtu.be/hOW63iiScgQ?feature=shared
Skip to 5:32 here:
https://youtu.be/hOW63iiScgQ?feature=shared
All it does is use words to predict the next words. That's it.
The rest is just a net of checks, triggers, and rules. For instance, when it generates an image, it is because threshold of word associations having to do with image generation is crossed, which automatically triggers a separate model to create an image. The LLM itself only predicted plausible word associations and plugged those into a prompt of its own for another tool to use, then it served the results. It doesnt "know how to generate an image". It can't "look at an image". Everything it does in excess of "guessing the next word" is handled this way.
It accurately hallucinates which words come next based on the strength of the relationships between instances of those words from its training data. A layer that reads its output will trigger other tools such as ocr image readers, search engine macros, etc as the need for it is detected via word association. The model accredited with intelligence, that generates words in your language, does ONLY that.
Thank you for attending my Tod Talk.
The rest is just a net of checks, triggers, and rules. For instance, when it generates an image, it is because threshold of word associations having to do with image generation is crossed, which automatically triggers a separate model to create an image. The LLM itself only predicted plausible word associations and plugged those into a prompt of its own for another tool to use, then it served the results. It doesnt "know how to generate an image". It can't "look at an image". Everything it does in excess of "guessing the next word" is handled this way.
It accurately hallucinates which words come next based on the strength of the relationships between instances of those words from its training data. A layer that reads its output will trigger other tools such as ocr image readers, search engine macros, etc as the need for it is detected via word association. The model accredited with intelligence, that generates words in your language, does ONLY that.
Thank you for attending my Tod Talk.
Toast message