Microsoft Study Finds AI Makes Human Cognition “Atrophied and Unprepared”
-
The AI will make sense of our terrible grammar.
-
i think that being used to properly structure sentences is important for reasoning well.
i agree that the effects of books and writing were probably beneficial to the brain, although they might have atrophied the memory and something else. But im not sure about tv, radio, internet and AI.
-
The majority of grammar rules are arbitrary and unrelated to the expression of an idea. For example, does it really matter if you treat an inanimate object like a pencil as feminine or masculine? It's an object. Yet in Spanish/French/etc., there are grammar rules that define every inanimate object as being either feminine or masculine.
However, without a common grammar, it's impossible to communicate accurately. For that use case, AI functions as a language translator.
-
Ehhh yes and no. There’s prescriptive grammar (how it ought to be) and descriptive grammar (how it’s actually used within communities). This is where the ideas of code switching and such come in. You can certainly reason well in a Creole, if that’s what your community speaks and how you are taught, e.g. Belizean Creole.
-
yes, its very arbitrary, but these are sets of rules that you can use to structure your thoughts. language helps us reason. it doesnt matter that it is arbitrary. definitions in mathematics are very arbitrary, but they are a foundation we can lean on to reason about abstract ideas.
Being arbitrary is not a testament of uselessness.
Different languages, lead to different foundations for structuring ideas. But dominating at least one of those foundations can be very important cognitively. -
Thankfully the slop generated by copilot et al is absolutely useless dreck. I've had a significant number of tasks end up broken because someone chased a dream promised by Ai slop. "Sure, you can do that in python." "that's definitely how that tool works." etc.
-
So the pros are
- thing we have no way of knowing how it works, therefore no way of relying on it
- thing that helps you do something you then have to do anyway by yourself (if you want to learn something from generative model output you still need to fact check it)
- vague promise it will lead to anything useful in the future
-
Books became a bicycle for the brain. Imo, AI is the same. Skills such as structuring sentences into perfectly grammatically correct forms will atrophy in exchange for the focus to be on the idea.
"In the future all our thoughts will be filtered through phone keyboard next word suggestion, and this is a good thing!"
-
yes, i wasnt advocating you should know any specific grammar. and that distinction is a good point. I meant that learning a prescriptive grammar decently is an important tool for reasoning.
im not saying that descriptive grammars are bad, just defending that prescriptive grammars arent as useless as people seem to judge them. -
So, AI users exhibit a reduction in literally the one skill that the AI expects them to actually have?
I should probably go read that link and see if it's actual degradation or just selection.
-
Gendering an explicitly non gendered inanimate object helps structure your thoughts?
I'd argue that following those grammar rules damages your thoughts.
-
Spoiler alert : it was just a survey of the reported confidence of folk who admitted to using AI.
-
Also don't forget Europe in the 18th century and how reading was destroying the youth. German Wikipedia has a big ass entry on it.
https://de.m.wikipedia.org/wiki/Lesesucht
The US Wikipedia entry is just a blurb.
-
You don’t always have to know how it works to rely on it. Most people could not tell you how a computer works but they are able to do better work.
We can verify that it’s better in some tasks than people. E.g give doctors and the AI 1000 MRI scans of potential cancer patients and it can determine it more accurate than doctors. So there they already are a help.
It’s already used in advancing different fields, for example reading texts of ancient burned scrolls without opening the scroll since that would break them.
But also medicine creation etc.
-
You don’t always have to know how it works to rely on it. Most people could not tell you how a computer works but they are able to do better work.
We can verify that it’s better in some tasks than people. E.g give doctors and the AI 1000 MRI scans of potential cancer patients and it can determine it more accurate than doctors. So there they already are a help.
I'd really prefer if my doctor knew why they say I have cancer!
-
what do you mean by "explicitly non gendered"?
-
Calculators made mental math obsolete. GPS apps made people forget how to navigate on their own.
Maybe those are good innovations or not. Arguments can be made both ways, I guess.
But if AI causes critical thinking skills to atrophy, I think it's hard to argue that that's a good thing for humanity. Maybe the end game is that AI achieves sentience and takes over the world, but is benevolent, and takes care of us like beloved pets (humans are AI's best friend). Is that good? Idk
Or maybe this isn't a real issue and the study is flawed, or more realistically, my interpretation of the study is wrong because I only read the headline of this article and not the study itself?
Who knows?
-
Why is a pencil masculine? It's a pencil, not a person.
-
It's developed by the worst of us and taught by a bunch of shit it read on reddit. You're thinking it might be benevolent?
