WARNING ! AI rots your Brain

michaelkellett said:

(if you've been using ChatGPT a lot recently you may struggle.)

Unless we're careful this could be very detrimental to childhood development in their formative school years.  I wonder what effects this will have on the aging brain.  Will there be an increase in Alzheimer's and dementia?

This is a no-brainer. 

But what does AI have to say for itself on the topic? 

Calculators rotted our brains.

Search engines eliminated the need to learn.

AI eliminates the need to think.

At risk is human creativity.

Its interesting that the use of search engines still appeared to require the use of brain to a much greater degree than using AI - so it is possible that AI pushes brain rot to a new level.

MK

Using ChatGPT also requires the use of your brain to discern what information is valid and useful.  It can be a real brain teaser but also a horrible time sink if you're not careful.

Hehe warning accepted! : ) 

dang74 good point, although for a child they can be "forced" : ) to be treated differently, because adults have made the decision there are things that all kids must learn, even if when thery grow up they don't need them if they have a tool, e.g. maths without a calculator until they reach a certain age. Parents already limit kid's TV time, turn off the Internet and so on. It's unlikely parents or teachers will embrace AI for kids to use at a young age, but I agree there's a danger, because not all parents will be aware.

Back to the article, I agree with what's likely the underlying reality, that there's something to be said for being in the practice of stretching your brain, and if you don't do it, you will struggle to do it. 

I bet this is why some people don't like long holidays. And I've mentally associated learning from lectures with writing notes, and I'm stuck if someone is teaching me something and I don't have a pen and paper to scribble on (even if I don't read the notes ever again; it's how my brain is learning).

It does feel to me that the experiment only reveals some things we would intuitively know.

If the AI was replaced by someone people were told was an expert on the subject at hand ready for their use, would people use them or not, and would then the brain work any different?

Incidentally, it feels like it would almost be wrong not to ask someone who you knew was an expert on a subject more than yourself (i.e., if indeed they were an expert and you have reason to believe they were).

Or (to simulate early AI), if the AI was replaced by someone the participants were told was definitely a non-expert, would the participants initially engage their brain more and not trust them? And there is a danger (for at least a period of time, probably not permanent - I don't think the study showed that!) that if you've not done the work yourself, or now realize it's a mundane task that someone else or a machine can do, maybe the brain is no longer all that keen to want to be used for such tasks!

Just some more random examples; I can see it in myself with books; if I don't read fiction for pleasure for a while, then it takes me half a dozen pages before it clicks that I'm not absorbing and am only scanning, and I have to go back and re-read and get back into the 'mode' of actually enjoying the book.

Another example of not being able to use the brain as much, is that I have a friend who is frustrated no end by a customer who regularly contacts him about basics which they could have figured out for themselves, e.g. to try a second cable if the first one appears to not work. The customer has mentally decided that he doesn't 'get' electricals, and offloads the thinking to his supplier (and I can totally believe it conditions the brain on that topic that he now needs that reliance, leaving him genuinely stuck if the supplier doesn't answer the phone). Maybe to get to that stage maybe there was initially also some inconsideration of the supplier (or maybe mentally thinking they have paid for the support by buying the product), but now even if he feels uncomfortable, he kind of needs that support always.

Also, interestingly, I also think some people will go to endless lengths (and brain effort) to reduce future brain effort. I've been on a (non-engineering) 8-week course, and on week 7, a colleague was having password renewal problems. I logged them in on another PC using my password (something totally acceptable according to the instructor) but it was clear it was still bugging him, he would keep turning to the first machine to reattempt, even after the instructor had told them it was very unlikely that they would need to log in on the 8th week. I wondered why it bugged him, he was a smart guy, so was it just for the challenge, to refuse to be defeated by a machine? I think the real reason is much simpler: Even if on the rare chance that he had to log in one last time, he would have to rely on humans to help him, which would take that bit longer, and would be a reliance on human variability! He would instead rather spend a lot of brain effort trying to eliminate that, for an easier log-in on the off-chance he did need to log in next week. I suppose a test could be if the participants in that study had the choice of using their brain to write the essay from scratch, or connecting up a PC and then using AI or Google, how many of them would have spent untold effort trying to fix IT issues first?

I think that the essence of the cited report is that using ChatGPT does NOT engage brains in a sustainable or fruitful way  and is much worse at this than more traditional ways of learning including web searching.

MK

My personal experience prototyping with AI aid often feels like my brain is exploding. This is because, while beneficial in some respects, the process proves quite exhaustive, demanding constant redefinition of specifications, multiple corrections, and frequent rephrasing of prompts to prevent hallucinations, sometimes even requiring a complete reset of the interaction history. I've observed this to be mentally more taxing than developing from scratch, carrying a significant risk if the generated code isn't meticulously verified. Future AI iterations ideally need better context retention and more robust error detection to reduce such extensive manual oversight, especially as complex logic still heavily relies on human expertise.

At this moment, I think it is more useful to find the relevant APIs, and get example code of how they are used.
Trying to let it build a working (demo) application, by chatting: no.