I’m not saying that AI is not without many serious flaws, but your simplification is highly inaccurate. If AI was based on getting “lucky” it would not be a marketable product, but rather just a parlor trick. What it is actually is a computer that can search the web much faster than you could, and provides results based on that search. It is not 100% accurate, but using a direct web search or more advanced model is pretty damn close for most purposes (especially something simple like the case in the article). It’s like calling someone who uses a calculator lazy. Ridiculous thing to say.
That’s a misrepresentation of what LLMs do. You feed them a fuckton of data and they, to oversimplify it a bit, put these concepts in a multi-dimensional map. Then based on input, it can give you an estimation of an output by referencing said map. It doesn’t search for anything, it’s just mathematics.
It’s particularly easy to demonstrate with image models, where you could take two separate concepts, like say “eskimo dog” and “daisy” and add them together.
When you query ChatGPT for something and it “searches” for it, it’s either fitted enough that it can reproduce a link directly, or it calls a script that performs a web search (likely using Bing) and compiles the result for you.
You could do the same, just using an actual search engine.
Hell, you could build your own “AI search engine” with an open weights model and a little bit of time.
It depends on the model and who made it, as well as what you are asking. If its a well known fact, like “who was president in 1992”, then it is math as you say, and could be wrong, but is more often right than not, but if it’s something more current and specific, like “what is the best Italian restaurant in my area” then it does in fact so the search for you, using google maps and reviews and other data.
They’re accurate enough for simple questions like “when was Bill Clinton President”? Go ahead and prove me wrong and ask that question to an AI, and show me one that gets it wrong.
“Accurate” and “accurate enough” have completely different meanings. Calculators are not “accurate enough”, they are accurate, and the idea that you’re conflating the two notions is exactly why LLMs are useless for most things people employ them for.
I’m not conflating the two notions. I have said that they are not completely accurate, but they are absolutely accurate enough. It is really very clear the experience of those who actually use AI, vs those who just regurgitate sensationalized headlines. If you think AI is literally “useless”, then you are not living in reality.
You are indeed conflating the two ideas, and I said “useless for most things they’re utilized for”, but if you quoted the entire sentence your argument would fall apart and you realized that.
I’m not saying that AI is not without many serious flaws, but your simplification is highly inaccurate. If AI was based on getting “lucky” it would not be a marketable product, but rather just a parlor trick. What it is actually is a computer that can search the web much faster than you could, and provides results based on that search. It is not 100% accurate, but using a direct web search or more advanced model is pretty damn close for most purposes (especially something simple like the case in the article). It’s like calling someone who uses a calculator lazy. Ridiculous thing to say.
That’s a misrepresentation of what LLMs do. You feed them a fuckton of data and they, to oversimplify it a bit, put these concepts in a multi-dimensional map. Then based on input, it can give you an estimation of an output by referencing said map. It doesn’t search for anything, it’s just mathematics.
It’s particularly easy to demonstrate with image models, where you could take two separate concepts, like say “eskimo dog” and “daisy” and add them together.
When you query ChatGPT for something and it “searches” for it, it’s either fitted enough that it can reproduce a link directly, or it calls a script that performs a web search (likely using Bing) and compiles the result for you.
You could do the same, just using an actual search engine.
Hell, you could build your own “AI search engine” with an open weights model and a little bit of time.
It depends on the model and who made it, as well as what you are asking. If its a well known fact, like “who was president in 1992”, then it is math as you say, and could be wrong, but is more often right than not, but if it’s something more current and specific, like “what is the best Italian restaurant in my area” then it does in fact so the search for you, using google maps and reviews and other data.
Calculators are accurate. LLMs are not, per your own admission.
They’re accurate enough for simple questions like “when was Bill Clinton President”? Go ahead and prove me wrong and ask that question to an AI, and show me one that gets it wrong.
“Accurate” and “accurate enough” have completely different meanings. Calculators are not “accurate enough”, they are accurate, and the idea that you’re conflating the two notions is exactly why LLMs are useless for most things people employ them for.
I’m not conflating the two notions. I have said that they are not completely accurate, but they are absolutely accurate enough. It is really very clear the experience of those who actually use AI, vs those who just regurgitate sensationalized headlines. If you think AI is literally “useless”, then you are not living in reality.
You are indeed conflating the two ideas, and I said “useless for most things they’re utilized for”, but if you quoted the entire sentence your argument would fall apart and you realized that.
How many R’s are in strawberry?
Did I say it was 100% accurate?
Found the glue on pizza enjoyer.
How many small rocks should I eat per day?
It depends. Are you a lizard or a bird? How long has it been since you last ate any?