MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline • Hacker News Board

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#261

Post by colincooke » Wed, Sep 03, 2025, 2:50 PM UTC

It is worth noting that this study was tbh pretty poorly performed from a psychology/neuroscience perspective and the neuro community was kind of roasting their results as uninterpretable.

Their trial design and interpretation of results is not properly done (i.e. they are making unfair comparison of LLM users to non-LLM users), so they can't really make the kind of claims they are making.

This would not stand up to peer review in it's current form.

I'm also saying this as someone who generally does believe these declines exist, but this is not the evidence it claims to be.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#262

Post by CuriouslyC » Wed, Sep 03, 2025, 2:50 PM UTC

blibble wrote: post #202

CuriouslyC wrote: post #189

Earlier quoted context omitted.

We're just moving up the abstraction ladder, like we did with compilers. I don't care about the individual lines of code, I care about architecture, code structure, rigorous automated e2e tests, contracts with comprehensive validation, etc. Rather than waste a bunch of time pouring over agent PRs I just make them jump over extremely high static/analytic hurdles that guarantee functionality, then my only job is to ide…

I can count the amount of times in my 20 year career that I've had to look at compiler generated assembly on one finger and I've never looked at the machine code produced by an assembler (other than when I wrote my own as a toy project) is the same true of LLM usage? absolutely not and it never will be, because it's not an abstraction

I like to think of it as a duck-typed abstraction. I have formal specs which guarantee certain things about my software, and the agent has to satisfy those (including performance, etc). Given that, the code the agent generates is essentially fungible in the manner of machine code.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#263

Post by MobiusHorizons » Wed, Sep 03, 2025, 2:51 PM UTC

mrits wrote: post #177

benterix wrote: post #173

Earlier quoted context omitted.

We are literally witnessing the skills split right in front of our eyes: (1) people who are able to understand the concepts deeply, build a mental model of it and implement them in code at any level, and (2) people who outsource it to a machine and slowly, slowly lose that capability. For now the difference between these two populations is not that pronounced yet but give it a couple of years.

I suppose the question is if we need to understand the concepts deeply. I'm not sure many of us did to begin with and we have shipped a lot of code.

I see a lot of junior engineers, or more senior engineers who are outside their areas of expertise try to prioritize making progress without taking the time to understand. They will copy examples very closely, following best practices blindly, or get someone else to make the critical design decisions. This can get you surprisingly far, but it’s also dangerous because the further past your understanding you are operating, the more you might have to learn all at once in order to fix something when it breaks. Debugging challenges all your assumptions, and if you don’t have models of what the pieces are and how they interact, it’s incredibly hard to start building them when something is already broken. Even then, some engineers don’t learn the models when debugging and resort to defensive and superstitious behaviors based on whatever solution they stumbled on last time. This is a pretty normal part of the learning process, but some engineers don’t seem to get past this stage. Some don’t even want to.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#264

Post by ramesh31 » Wed, Sep 03, 2025, 2:51 PM UTC

I think like a lot of people here, my posture towards AI usage over the last 2 years has gone from:

"Won't touch it, I'd never infect my codebase with whatever garbage that thing could output" -> ChatGPT for a small function here or there -> Cursor/Copilot style autocomplete -> Claude Code fully automating 90% of my tasks.

It felt like magic at first once reaching that last (current) point. In a lot of ways for certain things it still is. But it's becoming clearer and clearer that this will never be a silver bullet, and I'm ready to evolve further to "It's another tool in the toolbox to be applied judiciously when and where it makes sense, which it usually does not.". I've also come to greatly distrust anything an LLM says that isn't verified by a domain expert.

I've also felt a great amount of joy from my work go away over this time. Much as the artisans of old who were forced to sit back and supervise the automated machines taking over their craft churn out crappier versions of something faster. There's more to this than just being an old fart who doesn't want to change. We all got into this field for a reason, and a huge part of that reason is that it brings us joy. Without that joy we are going to burn out quickly, and quality is going to nosedive.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#265

Post by upcoming-sesame » Wed, Sep 03, 2025, 2:51 PM UTC

e3bc54b2 wrote: post #228

CuriouslyC wrote: post #189

Earlier quoted context omitted.

We're just moving up the abstraction ladder, like we did with compilers. I don't care about the individual lines of code, I care about architecture, code structure, rigorous automated e2e tests, contracts with comprehensive validation, etc. Rather than waste a bunch of time pouring over agent PRs I just make them jump over extremely high static/analytic hurdles that guarantee functionality, then my only job is to ide…

As the other comment said, LLMs are not an abstraction. An abstraction is a deterministic, pure function, than when given A always returns B. This allows the consumer to rely on the abstraction. This reliance frees up the consumer from having to implement the A->B, thus allowing it to move up the ladder. LLMs, by their very nature are probabilistic. Probabilistic is NOT deterministic. Which means the consumer is neve…

agree but does this distinction really make a difference ? I think the OP point is still valid

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#266

Post by Shank » Wed, Sep 03, 2025, 2:52 PM UTC

colincooke wrote: post #261

It is worth noting that this study was tbh pretty poorly performed from a psychology/neuroscience perspective and the neuro community was kind of roasting their results as uninterpretable. Their trial design and interpretation of results is not properly done (i.e. they are making unfair comparison of LLM users to non-LLM users), so they can't really make the kind of claims they are making. This would not stand up to…

> It is worth noting that this study was tbh pretty poorly performed from a psychology/neuroscience perspective and the neuro community was kind of roasting their results as uninterpretable.

Do you have links or citations to people saying these claims?

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#267

Post by epolanski » Wed, Sep 03, 2025, 2:53 PM UTC

I can't but think this has to be tied to _how_ AI is used.

I actively use AI to research, question and argue a lot, this pushes me to reason a lot more than I normally would.

Today's example: - recognize docs are missing for a feature - have AI explore the code to figure out what's happening - back and forth for ours trying to find how to document, rename, refactor, improve, write mermaid charts, stress over naming to be as simple as possible

The only step I'm doing less is the exploration/search one, because an LLM can process a lot more text than I can at the same time. But for every other step I am pushing myself to think more and more profoundly than I would without an LLM because gathering the same amount of information would've bene too exhausting to proceed with this.

Sure, it may have spared me to dig into mermaid too, for what is worth.

So yes, lose some, win others, albeit in reality no work would've been done at all without the LLM enabling it. I would've moved to another mundane task such as "update i18 formatting of date for swiss german customers".

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#268

Post by noboostforyou » Wed, Sep 03, 2025, 2:55 PM UTC

benterix wrote: post #173

goalieca wrote: post #164

Anecdote here, but when I was in grad school, I was talking to a PhD student i respected a lot. Whenever he read a paper, he would try to write the code out and get it working. I would take a couple of months but he could whip it up in a few days. He explained to me that it was just practice and the more you practice the better you become. He not only coded things quickly, he started analyzing papers quicker too and…

We are literally witnessing the skills split right in front of our eyes: (1) people who are able to understand the concepts deeply, build a mental model of it and implement them in code at any level, and (2) people who outsource it to a machine and slowly, slowly lose that capability. For now the difference between these two populations is not that pronounced yet but give it a couple of years.

And it's almost the same exact trend in skill atrophy we saw between millennial and gen z/alpha where most kids these days cannot do basic computer troubleshooting even if they've owned a computer/smartphone practically their entire lives.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#269

Post by aprilthird2021 » Wed, Sep 03, 2025, 2:55 PM UTC

elAhmo wrote: post #231

aprilthird2021 wrote: post #204

Earlier quoted context omitted.

> We're just moving up the abstraction ladder, like we did with compilers. We're not because you have to still check every outputted code. You didn't have to check every compilation step of a compiler. It was testable actual code, not non-deterministic output from English language input

I would bet a significant amount of money that many LLM users don’t check the output. And as tools improve, this will only increase. The number of users actually checking the output of a compiler is nonexistent. You just trust it. LLMs are moving that direction, whether we like it or not

Of course they don't. That's why things like the NX breach happen. That's also why they don't learn anything when they use these tools and their brains stagnate.

Re: MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline

#270

Post by pessimizer » Wed, Sep 03, 2025, 2:58 PM UTC

Taek wrote: post #97

IAmBroom wrote: post #91

Earlier quoted context omitted.

Absolutely true. Also, domesticated dogs show indications of lower intelligence and memory than wolves. They don't have to plan complex strategies to find and kill food, anymore.

The difference between us and dogs is that we DO still need to make a salary. Dogs live in a lap of luxury where their needs are guaranteed to be handled. But humans need jobs, and jobs need to capture value from society. So we do actually still have to stay sharp, whatever form "sharp" takes.

You and dogs have the same job, which is to please the boss. The boss then takes care of you like a child, either with a paycheck (with which you can pay servants to supply your earthly needs), or directly if you're a dog and lack both thumbs and pockets to hold a wallet or a phone. A domestic dog would die left alone in a forest, about two or three weeks after you would.

If you're an entrepreneur, your job is to please the customer and to squeeze your vendors and employees. You still take little to no part in directly taking care of yourself, except as a hobby. Unless you want to be congratulated for wiping your own ass or lifting a fork to your mouth.