Brain activity much lower when using AI chatbots, MIT boffins find
theregister.com
theregister.com
at least more people get to see the word boffin today /s
She made a really good point: we're in the TikTok era.
After she said this, it made much more sense. The whole movie felt like "reels" stitched together, with no scene lasting longer than 5 minutes.
It makes me wonder if people today (especially younger people) would be able to watch a movie like the Godfather? I have doubts.
Honestly, I'm terrified of where we are heading as a species.
I felt the same way about the first hour of the new mission impossible movie. It was so annoying to me that I left after that, so maybe it got better in hour two.
This comment is spot on. It felt like a mix between a short, and ad for... Generation tiktok
Last time I watched the movie 'The Irishman' with family, they all seem pretty tuned out by 15 mins.
It does feel like like addiction to reels/tiktok does have its toll. For that matter, due to Twitter exposure, I do have problems reading books.
When I probed ChatGPT a little, it tells me writing with a real pen and paper is now a super power as it lets you move slowly, pause, think etc.
Yes, exactly. So we offload mental tasks to AI, and avoid things that force us to concentrate for longer than 5 minutes. Cognitive decline is almost guaranteed.
I won't use AI for emails or creative content, but I'll for sure use it as a sounding board and as a trivial task completer. I bet my brain activity is lower using a calculator.
Nuance does not make for a flashy article title.
Is it safe to say that people are generally much worse at mental math nowadays due to calculators being ready available on every smartphone? I think it is reasonable to assume that's true. Math is more accessible thanks to tools now, but people are worse at it in general?
If that's true, that we became worse at mental math thanks to readily available tools, then what does it suggest about a readily available tool that "thinks" for us?
The same way we used to caution devs not to just copy and paste from Stack Overflow, we now have to repeat with AI... Except AI is being pushed to literally take over that whole workflow!
At least with Stack Overflow you had to identify where in the code to paste it!
Edit:
A further thought in Math. I suppose there is a difference between "Math" and "Calculation". Likely due to modern education people do know (at least vaguely) more about Math than before. Maybe the skill of actually performing calculations is not that valuable as long as you have some baseline understanding of the underlying Math
Still the problem with AI tools is they promise to do the Calculation and also understand the Math for you. To calculate an angle you need to know enough Trig to figure out which sides you need to measure and which function you need to use on the Calculator. With AI, in theory, you just ask it what it needs to calculate that angle, it tells you, you measure, input the numbers, and it gives you the angle back. No Human understanding of anything required. A total loss imo
You can view it as freeing our cognitive abilities for other things
I view it as never developing cognitive abilities because we never had to actually use them
People can still do math in their head (and on paper) despite the existence of calculators. They can still dig holes by hand despite the existence of shovels. And they can still write a for loop despite the existence of AI.
Because in the past we recognized the value of teaching people that way first
I don't know about you but when I was growing up we were not allowed to use calculators in class or on tests for years until we started to need them for functions like trig and graphing
The question is whether a learned skill is “much worse” in the presence of a tool. You seem to be agreeing that it isn’t (or at least doesn’t have to be), but that danger lies in just never learning the skill.
Yes, which is what I think AI contributes to
Especially with reckless ideas like "vibe coding" going around
"Vibe coding" is primarily being done by people who already weren't learning to program. If programming was taught as ubiquitously and early as arithmetic, they either wouldn't be vibe coding or they were never going to program in the first place because for whatever reason they were unable or unwilling to develop the skill. So you could see this as a loss of potential, but I suspect it is far more of a democratization of programming.
If we're looking at parallels, we should be looking at WYSIWYG webpage creation tools rather than calculators. Did that destroy knowledge of HTML and CSS?
Now apply that to LLM used to writing, research, reading comprehension, summarizing, debating, and so on, in the same school system
That's the point of learning arithmetic first without calculators.
Turns out AI could put us in a chinese room.
Well, that's one of the more terrifying things I've heard this week...
Trapping us all in a simulation does seem like the same general idea
I don't think that we should not use tools to offload tedious work
However there must be tradeoffs, right?
Take physical fitness for an example. We built machines to automate a ton of physical labour for us. Many of us no longer work long hours in hard labour jobs. A single excavator driver can move more earth faster than a team of people with shovels could even hope to
How's our physical fitness as a society these days? Yes part of that problem is that we eat terrible foods, but still. The sedentary lifestyle plays its part in the obesity epidemic too
I think it's reasonable to speculate that "we no longer need to think much, just pilot the AI" will similarly decrease our mental capabilities
Nope. The answer is government and money.
As someone who had a father that taught students and now has kids (that I teach myself, and help them go to school) I know the direct answer: teachers. Also: mental math did not drop off a cliff when calculators appeared, nor when they were accepted. It slowly slid downward. More than that: it peaked. It's not like the people who fought in WW2 were good at mental math either. So between 1960 and 2000 something happened to greatly push mental math ... and now to destroy that ability.
A long-term perspective on the teacher profession also taught me why. In the 70s few people were educated. Why would you? And worse: we were all working in industry. That's where the money was ... but not if you were educated (which, let's be honest, tends to instill less willingness to do repetitive manual labor). The limiting factor of growing industry was: geometry, calculus and fast mental math. That's what you require to make good plans and good use of factories to turn metal into products.
A lot of smart, or at least educated, people were lured into government positions in the 70s by the previous economic crisis, and the new growth market of industry were pushing and pushing and pushing to teach math, from as young as possible. They taught themselves math because they started teaching careers and the money in these was hard to beat. Industry paid for advanced math courses for teachers. In the 70s teaching positions easily matched industry wages, except you mostly worked at a desk, and especially the pension plans were sweet. Teachers got free math course, paid for by industry, to teach themselves math because that got you better positions, making a bit more per month.
Of course, the government lied. Here's the HUGE caveat in a government contract: A government can simply vote in a law or even regulation that "overrides the contract", an therefore cannot be counted on to keep long-term promises even when there is a signature on a paper. And while unions help slow the destruction, they cannot stop the slide. So starting after the crisis at the end of the 1980s governments lowered teacher pay, even for existing people (which is where government power comes in: private sector firms are not allowed to do this). Which then also meant they accepted a much lower quality of people into the teaching profession, as they couldn't compete anymore, the government simply did not care, and existing teachers got discouraged. Now, mind you, "quality" refers to their academic ability, as certainly a part of teacher pay was (and is) that you work all day in a position where you socially dominate everyone around you, which of course led to abuses of that position as well. And since it's not like lower academic quality people are not abusing their positions as teachers, that part has stayed (just look up developmental psychology papers: almost all child abuse happens in schools, and what has the government done about that? Well: CPS was legally forbidden from investigating schools, or teachers. Only parents get investigated and even blamed for school problems (that's in theory, in practice of course it's mostly the least powerful that CPS really attacks: the kids themselves))
This happens via brain connections and brain activity. The good times means less tribulations and the weak men simply havent formed the brain pathways.
AI makes for 'good times' and the easy 'low brain engagement' is the identified problem.
These scanned folks likely are using AI wrong.
Never ask AI "summarize this book" for me. You're missing out on reading the book. What if you read the book and then ask "What did this book miss?" or "give me ideas on what i can do to expand on the ideas in this book"
The way i look at it, dont 'reduce' only 'expand'
The way I found this, back in the day video games were awesome and then an EXPANSION was added that expanded the game. So when working on my open source projects, I was always asking 'whats the next expansion?"
Something that i love asking ai. "Here's my project and what it does. What are the next big ideas that can be added."
This is an aphorism that mostly doesn't bear out historically. Good times tended to make people wealthy, and wealthy people tended to be stronger, not weaker.
I've thought of that aphorism from time to time when I encounter people on vacation here in Mexico, let's say, who seem to become a wreck when some small thing doesn't go as they'd liked.
Like the kitchen is outside next to the pool, so there's no A/C and how dreadful is that. And the boy at the cenote charged them 60 pesos ($3) while a sign says that resident Mexicans only pay 30 pesos and that's unfair for someone who has 10x the wage of the resident Mexicans. And someone had a chicken on the bus going to Guadalajara. And now they're going to grumble with their spouse for six hours over it. And there was an iguana in their shower this morning so now they're getting management involved.
I'm thinking this is a weak, pathetic man crippled by a life too comfortable.
Not all adversity should be endured, that's true. Like dealing with an evil manager at work.
But when someone finds a line of leaf cutter ants in front of their hotel door or a shower iguana to be something worth bickering over with management despite booking a trip to the subtropics, they've been rotted by comfortable times.
I wholly agree with the aphorism and it's a defeatable law; but perhaps it's not exactly clear who they mean. It's not about father-son type situation. It's completely unclear what 'strong' they even mean. Not to mention the other law that's counter to this in that who is defining weak vs strong?
My argument against the aphorism is the cycles can be happening out of sync. It's not like 100% of men are weak in any era. It's not that you cant intentionally break the cycle.
Rockefellers, Buffet, Bill Gates, Elon Musk, and andrew Carnegie are all perfect examples of breaking out of the cycle.
Perhaps we need to rewrite the saying:
Difficult times forge resilient people, resilient people build prosperous societies, prosperity breeds complacency and dependence, complacency and dependence usher in difficult times.
Also historically, difficult times don't breed resilient people, the dark age was a millennium long.
Not only can future development be inferred, you can see the pattern in history. The consequences of weak men.
Reconstruction -> third great awakening -> world war 1 -> roaring 20s -> world war 2 -> golden age of capitalism -> 4th great awakening -> culture wars, 9/11, tech bubbles -> great recession, covid, etc
Or to look at what we're in right now.
Glorious revolution 1701 + 75 years = age of revolution 1776 + 84 years = civil war 1860 + 79 years = world war 2 1939 + 80 years = 2019
The cycle is 4 movements in 80 years adjusting every 20 years. Hence the 4 parts of strong men create good times.
>Also historically, difficult times don't breed resilient people, the dark age was a millennium long.
This is where you're confused. This isn't a situation where the USA is doing great and 100% of people are strong, then create a generation of 100% weak people.
During the civil war, there was weak and strong people. then during reconstruction there was weak and strong people.
This feels like victim blaming.
If the tool can be so easily misused that the participants in this study "used it wrong" by default, then were are the safeguards? How are these companies nudging the users in the "right" direction?
The way I see it, if there's nothing in the system telling you that you're using it "wrong" then you're using it exactly as intended.
How come nobody is telling them to use the other side? If the hammer doesnt have instructions on the side to say to use the hammer side, then it's being used exactly as intended.
What I've never seen in my life is a random group of 9 people (per the study) where all of them consistently use the wrong side of the hammer repeatedly across four sessions.
I love the use of the word "actually" there, as if anybody were under the illusion that using LLMs makes individual humans smarter.
I expect better from El Reg.
The Human problem with "the everything machine" is that it does "everything" and by and large Humans really do need to do something while we're alive
Because anyone could do that.
To remain competitive, humans will have to keep using their cognitive functions.
I cannot imagine how miserable it will be trying to keep up with the AI slop firehose
If the human doesn't do it, no one does.
So great if the AI can help with the subtasks, bad if people assume the big problems will solve itself.
Because AI can’t do it. It does not understand anything and probably won’t before a very long time.
EVE ponytail length
e.g. "UK gov asks university boffins to pinpoint cyber growth areas where it should splash cash"
It looks like the request from 2023 fell on deaf ears.
I do not have many options for discussions anymore. This is for various reasons.
Some of the dialogue I've experienced through various LLM systems has been far more intriguing and thought provoking than 90 percent of my human conversations. I distinctly recall thinking how great it was to have something to substitute the former roles of bygone friends, some of who were both highly educated and intelligent.
Some of my more enjoyable discussions have been in the realm of guerilla physics. So one could easily enough argue that my mind was wasting away while being cheaply entertained by bullshit. But other subjects have been equally satisfying and quite rewarding. Some credit may be due to very meticulous prompt construction. Or perhaps the simple earnestness of my inquiries. If my time spent in these sessions has reduced my cognitive activity, it is miraculous that I can function so well in a coma around most actual people, who do verily inspire one -- a coma, that is.
A deep one.
Using a bow engages more faculties, both physical and mental than clobbering someone on the head but it's also a more effective tool. Using a debugger is both a difficult skill but also useful. You grow, and have to increase your aptitude , with the tool use. An exercise bike and meditation practice are harder than a sleeping pill and ozempic.
Chatbots are more in the category of the 400 pound guy driving through wallmart on his electric scooter. Whatever use it has, it almost certainly diminishes if not totally atrophies their ability to move. Given that the modern framing of technology has changed to "what does it do for me" rather than "what does it do to me", it's not surprising that people are largely not capable to understand the difference.
-- Nothing like a 100% accurate statement getting downvotes because people hate AI
In the past, computers, calculators etc. have reduced the need for type 1 thinking, but I believe LLMs are increasingly replacing (or emulating / mimicking) type 2 thinking which - again entirely in my view - is more harmful to individuals and society as a whole.