When I'm really using AI, my mind is pushed to its very limits. I'm forced to maintain context that is much more complex than anything I had to keep in working memory pre-AI. But it also feels easier because you don't have to do nearly as much thinking to get every given task done. So maybe I get lazier, not in how much I accomplish, but in how much effort I put forth. So if my previous working intensity applied with AI would let me finish 10x as much work, now I'm content with exerting half as much effort and getting 5x as much work done as my pre-AI self.

I’d encourage folks to listen to this podcast[1] or read the transcript which is done by two incredibly respected people, Dr. Cat Hicks, a psychologist who studies software teams, and Dr. Ashley Juavinett, who is a practicing and teaching neuroscientist. They note the many flaws in the study and discuss what actually good brain research would look like.

1: https://www.changetechnically.fyi/2396236/episodes/17378968-...

In the people around me I’ve observed:

AI solves the 2-sigma problem when used correctly.

AI is extremely neurodegenerative when used incorrectly.

The people using it as a research assistant to discover quality sources they can dive into, and as a tutor while working through those resources, are getting smarter.

The people using it as an “oracle made from magic talking sand” are getting dumber.

To be fair, the same thing is true of the web in general, but not to the extreme I’ve been seeing with AI.

I’m predicting the bell curve of IQ is going to flatten quite a bit over the next decade, as people shift two sigma in both directions.

obviously obvious caveats like, intentional use is good, lazy use is bad, etc.

I've found it both helpful and dangerous, it's great for expanding scope obviously, greater search engine.

But I've also significantly noticed further some of the "harmful patterns" I guess that I would not have noticed about... myself? For example, AI is way too eager to "solve things" when given a prompt, even if you give it an abstract one. It's unable to take a step back and just.... think?

And hey, I notice that I do that too! Lol.

It's helped me realize more refined "stages" of thinking I guess, even beyond just "plan" and "solve".

But for sure a lot of the time I'm just lazy and ask AI to just "go do it" and turn off critical thinking, hoping that it can just 1 shot the problem instead of me breaking it down. Sometimes it genuinely works. Often it doesn't.

I think if I stay way more intentional with my thinking, I can use it to good use. Which will probably reduce AI usage - but it's the first principles of real critical thinking, not the usage of AI.

---

These kinds of studies remind me of when my parents told me "stop getting addicted to games" as a kid. Sure, anyone can observe effects, it takes real brains to really try and understand the first principles effects. Addiction went away in a flash once I understood the principles, lol.

I've been seeing people use LLMs to reply to people on Facebook. Like, they'll just be having a general discussion, and then reply as ChatGPT. I don't know if they think it makes them look smart; I think it has the complete opposite effect.

Not many people can perform mental arithmetic beyond single-digit numbers. Just plug it into a calculator...

We're at the point of people plugging their thoughts into an LLM and having it do the work for them... what's going to happen to thinking?

The future for humans worries me a lot. What evolutionary pressures will exist to keep us intelligent? We are already seeing IQ drop alarmingly across the world. Now AI comes in from the top rope with the steel chair?

https://www.ncbi.nlm.nih.gov/search/research-news/3283/

That would not surprise me at all given what I've observed in a couple of people who outsource the thinking part to LLMs. One of them has dropped at least 20 IQ points and went from being able to grasp complex concepts with ease to needing an LLM to confirm that indeed, 2+2=4 (only somewhat hyperbolic).

So users are more detached from their work? How does this correspond with cognitive decline? Wouldn’t it need to be cross referenced in other areas beside the task at hand? Seems a bit of a headline grabbing study to me. Personally I find thinking with an LLM helps me take a more structured and unbiased approach to my thought process

Does anyone have any suggestions for approaches they are taking to avoid the potential for this? Something I did recently in ChatGPT's 'Instructions' box (so far I have only used ChatGPT) is requesting it to "Make me think through the problem before just giving me the answer." and a few other similar notes.