The Side Effects of Overusing ChatGPT For Homework

To be fair people's brains where rotted way before Ai

So what this paper implies is that people with better cognitive discipline can use LLM as a tools more effectively than use it as a replacement. So maybe one of the way to measure the learning of a user is to stress test on their own answer regardless of if they use LLM or not, by forcing them to present their own work on the fly?

ChatGPT pls summarize this video

At 9:01 the authors tell journalists not to say “brain rot”, but the thumbnail says “brain rot 2.0”

It seems to me that the difference between LLM-to-brain and brain-to-LLM is a matter of who is "directing" the thinking process. In this study, both groups are (presumably) being introduced to novel subjects, meaning their opinions and ideas on the topic are largely unformed before entering the study. That means that the way you introduce yourself to this new information for the first time, whether it is through your own natural thinking skills, or that of an AI, would make a huge difference as to how you interact with the information as a whole. In the case of the LLM being your introduction, you aren't coming to conclusions on your own, you aren't going through that process of trial and error and discovery. Instead, you're being spoon-fed ideas and conclusions by the LLM (which seem credible enough and result in you adopting in some or all of those outputs as your own thoughts and opinions). You're not the person giving the speech, you're the audience listening to the speaker. In comparison, the brain-first group is forced to discover and form their own thoughts on the subject through research, which leverages each person's unique set of life experiences which serve as the background context to the way they think and perceive. This probably creates a far stronger connection between many parts of the brain, combining past and present memories to form a structurally sound understanding of the new concepts. Essentially, this approach creates true "ownership" of the ideas, whereas in the LLM-first group, they are merely "borrowing" somebody else's ideas. Once you've solidified a frame of understanding around a subject, even if you introduce LLMs as an assistant at a later time, now the person has the ideas pre-formed meaning they can act as the chef in the kitchen, not as the customer being fed (the 1 week break probably helps with this, actually, because it gives the students time to process the information in the background and really solidify and encode their understanding into their deeper, more static parts of their brain). I think what we can take away from this study is how absolutely crucial it is that we practice non-assisted, non-LLM thinking times many times throughout the day. If you're going to do research on a subject, you should first do the research un-assisted and try to explore and discover and support your curiosity, and then afterwards, you might be able to leverage an LLM to further expand on your research, with you directing the LLM's focus into the areas you are interested in (so, using it as a tool, not a crutch). It's probably also important that you continuously cycle between assisted thinking and non-assisted thinking, like even creating a cadence where you do your own research, leverage an AI to expand on what you researched, then use that information to further dive deeper with your own research, return back to AI, so on. Personally, I have never used LLMs (besides toying around with them when they first came out) and will never use them because I value my own ability to think, but I know that the cat is out of the bag so the best thing for all of us is to promote healthy ways of using it, not pretending it doesn't exist altogether.

Sample size of 18 and 9 is so inadequate that I wouldn't conclude anything at all from that.

Master AI agents now using HubSpot's FREE resource! https://clickhubspot.com/3a9164

The paper did not show that using the AI left you worse off than before, as 'rot' would indicate. It shows a gap of growth of something beneficial.

Problem with that study is that it’s only 18 people used ,that’s not enough data to produce a statistically significant result.

Cloud man, I think that lingering feeling is because there's a ghost cat in your house.