It’s always nice to meet people who have the same outlook on AI as myself, but there are still those that argue that “consumption itself has a huge impact” when talking about the immense environmental damage of AI. In those moments I find it useful to talk about the stupidity of ChatGPT and Arthur Perret does this especially well in this article: https://www.arthurperret.fr/blog/2024-11-14-student-guide-not-writing-with-chatgpt.html
=> More informations about this toot | More toots from dasha@post.lurk.org
personal favorite from the article: “ChatGPT is a human conversation simulator, not an information system or a knowledge base. It has no understanding of anything: it only outputs plausible responses.”
=> More informations about this toot | More toots from dasha@post.lurk.org
@dasha as much as I do agree with the ethical concerns (as always), I usually don’t find the appeal to the stupidity of chatbots very genuine. It does come across as gatekeeping, especially when followed by sending students away to libraries, professors and / or other institutions that might stall the learning process or make it more frustrating. And this is on the basis that AI doesn’t “understand”. The understanding is a concept that doesn’t have a single explication, and what chatbots appear to prove is, rather than that they are capable of understanding, that the solely-humanly attributable intelligence might not have as much value as our sense of exceptional importance as a species would suggest.
Also pointing to comprehension capabilities of AI seems to be missing the point here. I’d argue that rather than asking whether chatgpt can “understand” a topic we should ask whether it can help someone with their intended goal, that when it comes to learning, surely involves understanding something. And on this basis, you don’t have to look far for the accounts of its value for learning or research, among many other uses.
One more issue I have with this text: reducing a technology down to its mode of function (predicting next token, generating most plausible text, etc) is sort of a functional essentialism that disregards what a thing emergently does. To fully account for something, we need to go beyond pure mechanics, to get a broader perspective that includes for example how does it actually unfolds in realistic context.
I would recommend this autoethnographic research article as a nice piece of counterweight that also reflects my experience with chatbots pretty well. This experience in summary is that they cannot be used uncritically and in isolation, but when some caution is applied, they can often help you get a grasp of new things pretty damn efficiently. And this is only one of its many uses in learning.
https://link.springer.com/article/10.1186/s41239-023-00404-7
=> More informations about this toot | More toots from perpetualbeta@tldr.nettime.org
@perpetualbeta I totally get the point about gatekeeping. With any new technology there’s always a fine balance when criticizing it, but I don’t really see how suggesting that students should turn to professors or libraries would stall the learning process? Isn’t that what those people/ressources are for? From talking to students my understanding is that using ChatGPT seems to eliminate the learning process altogether by just spewing out the (often incorrect) answers.
=> More informations about this toot | More toots from dasha@post.lurk.org
@dasha well would be great if the libraries were always available and up to date, resources at the reach of the hand, professors not overloaded with responsibilities, and students determined enough, and given enough time to go through a lengthy process. In reality it’s hard to check these boxes, and a chatbot often turns out to be a viable alternative that although indeed does make mistakes, does not easily get impatient, does not judge, answers on command, and therefore helps getting in the zone. Also, while hallucinating sources, often not being able to connect the dots insightfully, it generally does provide solid ground for kicking research off and helps overcoming blocks. Also it excels in fulfilling pointless tasks, and writing uninspiring reports.
As often, I feel like if we find something is wrong about a technology we could ask what are the incentives for people to use it. Just deeming it stupid is not really gonna help move forward with any agenda if the very reason for why it finds so many enthusiasts has nothing to do with how it works, but rather with what kind of things we want to or have to do, that this thing makes us capable of doing.
That being said I’m far from falling for chatbot frenzy, I would not claim it’s intelligent, or say we should continue this direction. I wish we would be exploring ways of doing technology that augments our experience, rather than stripping us of it. But I do recognise that perhaps we are conflating a cause with effect and if we don’t like that people use chatbots, maybe would make sense to think of why they really do it, instead of distancing ourselves from their lived experience of others through mocking and ridiculising based on theoretical divagations.
=> More informations about this toot | More toots from perpetualbeta@tldr.nettime.org
@perpetualbeta I totally agree that often library ressources are lacking and professors overworked and that more money should be going into education. But I also know that not to be the only reason kids use ChatGPT. I can give a personal example of teaching in an art school recently where students were using it unprompted and also for tasks where it was completely unnecessary, all the while i was one teacher for 8 students and was able to provide personalized help.
=> More informations about this toot | More toots from dasha@post.lurk.org
@perpetualbeta I think unfortunately the allure of doing things faster and not having to think twice about it is too tempting! And as the article is talking about writing - I think that’s one of the educational activities that would hurt from being outsourced to a chatbot the most. It’s only through practice of writing that we develop critical thinking abilities and are able to learn to make better arguments (i struggled with writing a lot, so this really hits home)
=> More informations about this toot | More toots from dasha@post.lurk.org
@perpetualbeta As for the stupidity argument, maybe you’re right that one should put worth a more elaborate argument, but the post.lurk server only allows for 500 characters, so I’ll be calling chatbots stupid, because while they might be very complex, they’re ultimately very simplistic!
=> More informations about this toot | More toots from dasha@post.lurk.org This content has been proxied by September (3851b).Proxy Information
text/gemini