71–80 of 589 posts
A few things to note. 1. This is arxiv - before publication or peer review. Grain of salt.[0] 2. 18 participants per cohort 3. 54 participants total Given the low N and the likelihood that this is drawn from 18-22 year olds attending MIT, one should expect an uphill battle for replication and for generalizability. Further, they are brain scanning during the experiment, which is an uncomfortable/out-of-the-norm experi…
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.
Everyone is different. I don’t have a good grasp on the distribution of HN readers these days but I know for myself as a heavy user of LLMs, I am not sold on this for myself. I am asking more questions than ever. I use it for proof reading and editing. But I can see the risk as a software engineer. I really appreciate tools like cursor, I give it bite size chunks and review. Using tools like Claude code though. It be…
> I am asking more questions than ever. Wouldn't that be the expected result here? Less knowledge, more questions?
When I use LLMs, it’s less about patching holes in my memory and more about taking an idea a few steps further than I otherwise might. For me it’s expanding the surface area of inquiry, not shrinking it. If the study’s thesis were true in my case, I’d expect to be less curious, not more.
Now that said I also have a healthy dose of skepticism for all output but I find for the general case I can at least explore my thoughts further than what I may have done in the past.
No, vibe science is not as powerful as to be able to determine "long-term cognitive harm", especially when such "technical wonders" as "measurable through EEG brain scans." are used. > 83.3% of LLM users were unable to quote even one sentence from the essay they had just written Not sure why you need to wire EEG up, it's pretty obvious that they simply did _not_ write the essay, LLM did it for them, and likely didn't…
The idea that I would say 'write an essay on X' and then never look at the output is kind of wild. I guess that's vibe writing instead of vibe coding.
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/
Why does it matter? Some will become Eloy and some Trogs.
>the gentle, childlike Eloi and the subterranean, predatory Morlocks.
Seems like a nice metaphor for the current two political parties we are provided with.
Everyone is different. I don’t have a good grasp on the distribution of HN readers these days but I know for myself as a heavy user of LLMs, I am not sold on this for myself. I am asking more questions than ever. I use it for proof reading and editing. But I can see the risk as a software engineer. I really appreciate tools like cursor, I give it bite size chunks and review. Using tools like Claude code though. It be…
When writing was invented, societies started depending on long form memorization less, which is a cognitive "decline". When calculators were invented, societies started depending on mental math less, which is a cognitive "decline".
I'm sure LLMs are doing the same thing. People aren't getting dumber, they are just outsourcing tasks more, so that their brains spend more time on the tasks that can't be outsourced.
Everyone is different. I don’t have a good grasp on the distribution of HN readers these days but I know for myself as a heavy user of LLMs, I am not sold on this for myself. I am asking more questions than ever. I use it for proof reading and editing. But I can see the risk as a software engineer. I really appreciate tools like cursor, I give it bite size chunks and review. Using tools like Claude code though. It be…
> I am asking more questions than ever. I don't have a dog in this fight, but "asking more questions" could be evidence of cognitive decline if you're having to ask more questions than ever! It's easy to twist evidence to fit biases, which is why I'd hold judgement to better evidence comes through.
If I’m constantly asking “what does this mean again?” that would signal decline. But if I’m asking “what if I combine this with X?” or “what are the tradeoffs of Y?” that feels like the opposite: more engagement, not less.
That’s why I’m skeptical of blanket claims from one study, the lived experience doesn’t map so cleanly.
A few things to note. 1. This is arxiv - before publication or peer review. Grain of salt.[0] 2. 18 participants per cohort 3. 54 participants total Given the low N and the likelihood that this is drawn from 18-22 year olds attending MIT, one should expect an uphill battle for replication and for generalizability. Further, they are brain scanning during the experiment, which is an uncomfortable/out-of-the-norm experi…
I feel like saying papers pre peer review should be taken with a grain of salt should be stopped. Peer review is not some idealistic scientific endeavour it often leads to bullshit comments, slows down release, is free work for companies that have massive profit margins, etc. From my experience publishing 30+ papers I have received as many bad or useless comments as I have good ones. We should at least default to ope…
Absolutely not. I am an advocate for peer review, warts and all, and find that it has significant value. From a personal perspective, peer review has improved or shot down 100% of the papers that I have worked on -- which to me indicates its value to ensure good ideas with merit make it through. Papers I've reviewed are similarly improved -- no one knows everything and its helpful to have others with knowledge add their voice, even when the reviewers also add cranky items.[0] I would grant that it isn't a perfect process (some reviewers, editors are bad, some steal ideas) -- but that is why the marketplace of ideas exists across journals.
> Science should become a marketplace of ideas.
This already happens. The scholarly sphere is the savanna when it comes to resources -- it looks verdant and green but it is highly resource constrained. A shitty idea will get ripped apart unless it comes from an elephant -- and even then it can be torn to shreds.
That it happens behind paywalls is a huge problem, and the incentive structures need to be changed for that. But unless we want blatant charlatanism running rampant, you want quality checks.
[0] https://x.com/JustinWolfers/status/591280547898462209?lang=e... if a car were a manuscript
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.