Comment by dumberquestions
7 hours ago
The drop from 2018 to 2022 is as big as the one from 2022 to 2026, so it's not obvious whether AI had a role at all, although it's highly plausible.
I suspect the bigger culprit is the optimized monetization of human attention, because pre-algorithmic social media doesn't seem to be as destructive. The fact that science wasn't as affected as math and reading (both of which rely more on attention/practice than rote memorization) somewhat supports this.
Between 2018 and 2022, we saw the widespread adoption of social media and the rapid digitalization of education. Focusing on these key shifts, it seems clear to me that any kind of electronic learning has had a negative impact on education :p
Essentially every high schooler had an active Facebook account and essentially every school paper was written on a word processor by, like, 2010. By 2018 even Instagram is starting to be played out. I would be very surprised to see a material increase in the share of US or western students using social media over the 2018-2022 period.
You say this as if 2010 Facebook was comparable to 2022 Instagram/TikTok.
5 replies →
I'd say the main difference is that you couldn't take a device with good media capabilities, good and cheap-ish networking and lasting battery everywhere. Laptops, netbooks, tablets were a tad too bulky, lasted only a few hours unplugged and mainly wifi, and mobile internet was not all that great.
I don't know for kids specifically but average user time: 2012 1:30 2016 2:08 2019-now ~2:30
If child patterns match the average, an average 8th grader from now has accumulated something like 30% more than one in 2018.
If only we knew what could possible have happened right in that time period to really throw a wrench in the works of the education system...
4 replies →
I worked for an education tech company in the 90s. Seemed to clear to me even then that computers in the classroom were not going to be a positive influence. The way things have gone since then, it is no wonder that kids’ brains have turned into pudding.
I think that computers in the classroom can have a tremendous positive influence in specific areas when see with correct guidance. I think where we went wrong is replacing the classroom with ChromeOS and iPads.
3 replies →
I did tons of electronic learning as far back as the 80s and 90s, and still do.
It’s algorithmic social media. There’s loads of both direct causal and indirect evidence that algorithmic attention hacking has made everything worse: mental health, learning, socialization, motivation, productivity.
Short form content makes it even worse. It’s an endless barrage of short pointless tidbits with no connection, no plot, no larger theme, and often little to no content in the actual shorts. It makes trash TV look like high art from a lost golden age.
I don't know why this is even a tiny bit controversial. The idea that the media you consume in formative years has a huge effect on mental development seems extremely obvious.
I grew up in the 90s. Its not like all our media was Tolstoy. But for my cohort it was very normal to read young adult literature - which meant books with 100s of pages, or video games that were elaborate, challenging, and demanded imagination and focus. The shortest media we consumed was 30 minute sitcoms. We got aesthetics, narrative structure, problem solving, focus - and maybe even some semblance of values.
Today, and I can hardly believe it, it really seems to be the case that the average teenager has read zero books cover to cover, or even fully watched a movie as difficult as ET. What she has done, is consumed hundreds of hours of twitchy bizarre quasi-algorithmic video content through a blaring handheld terminal thats held more intimately than any friend or pet. Far from being content made with artistic integrity or true dedication to the fostering of the young, this stuff is more akin to direct neural stimulation.
What I don't understand, is why anyone would think that such a shift would result in anything other than marked mental deterioration.
1 reply →
Another modern problem with education is that every student has a smartphone now, which they bring to the class, then they get distracted by it.
"Modern" education isn't modern at all. It's a leftover from industrialization, aiming at producing easily controllable worker drones.
"Elites" fearing the "plebs" to question them on things they have no good answers for is what constrains active development of actual modern scientific concepts of pedagogy taking hold. You only get to research what is being paid for.
Look at the, pretty obvious, natural human child-mode of learning: playing games.
Just not escapist nonsense games. Games that model relevant aspects of reality in an engaging way. Where you can fail without real-life consequences and learn from it while enjoying yourself.
11 replies →
This is highly optional. You can lock phones up at the start of the school day or at the start of class
4 replies →
Maybe more and more teachers care less and less about training students to take tests, and care more about actually educating students in the subject matter. So maybe the plunging test scores are are simple as “students are worse at taking tests.”
4 replies →
I do not think this is right; Kahn Academy, MitX, opencourseware etc are treasures.
If they are used, sure.
What are you talking about?
Between 2018 and 2022, we had a global pandemic hit!
The only reason we were "digitalizing education" is because millions of people were getting sick and dying. Millions more lost their jobs.
Trying to blame any single aspect of what happened during that period, like moving to remote learning, is hopelessly myopic and blinkered.
We haven’t had a pandemic between 2022 and 2026. The 2018 trend continues. That strongly suggests a common effect across both samples.
1 reply →
If I understand you correctly, you are using Covid-19 as a means to justify any position (or broad classes of them) which is convenient to you. I say this because you're using it in a broad way, and making an appeal to specific prejudices.
You make it sounds like you're being sarcastic. But I would argue that ubiquitous digital interfaces demanding our attention is wreaking havoc on people's focus.
Hmm, did anything happen in 2020 to disrupt the entire education system?
But if the drop between 2018 and 2022 was AS LARGE as the drop between 2022 and 2026, this implies that something happened that was as large as a global pandemic for students' education happened in the last four years. I believe the OP is implying this could have been AI. However, something else could be happening, such as use of phones and other devices or another fundamental problem in the education system.
One can try to tease out effects by looking at per-country or per-state breakdowns. It seems to me that:
- there are lots of different long term trends that vary by country
- some seem to have started too long ago to be tick tock or similar
- some countries see a pandemic drop and recovery
- this implies to me that:
(I) the pandemic had a strong temporary effect on average educational attainment (though I have strong priors here), and
(II) other things like long-term trends in teaching approaches, new distractions for students, or availability of AI for ‘cheating’ also have an impact
See coverage of the recent test results in the FT here: https://www.ft.com/content/724f8ecb-1e2b-4c7f-bbb6-14239abd5...
- Students who entered the education system in the COVID years are still taking tests today. They didn’t automatically relearn everything.
- Schools have not gone back to pre-COVID practices. Online courses, use of digital tools, fewer tests, dropping SAT/ACT requirements, and most of all budget cuts in school districts – all are still happening today.
2 replies →
The trauma from lockdowns does not disappear overnight! It is compounding damage, that gets worse every year!
And we can not just change childrens habits, after we forced them to sit at home for 2 years with nothing but youtube! They missed key developmental phase!
9 replies →
Yeah, I am genuinely confused how the grandparent comment just glossed over it.
I believe that both COVID and LLMs are disrupting educational metrics. But in terms of which one did more short-term scale damage (i.e, within 4 years), I would wager it was COVID.
Wouldn’t there be recovery? Especially for the cohort that started school after the end of restrictions.
A continuous decline across both COVID and llms points towards a different causal relationship to me. I would be more likely to believe algorithmic social media or computerized teaching tools are to blame, but that is pure conjecture.
2 replies →
the event in 2020 would've only been the last straw in a series of bad policy in education. It might've brought forward this same outcome a few years in advance, but it isn't a root cause.
I saw extremely rapid computerisation of my kid's education around 2020. Properly thought out and edited text books were replaced with last minute rushed teachers notes on google docs and buggy apps.
That would have contributed to to the decline.
When I was a freshman in college taking CS courses, they were just rolling out a new Java curriculum. And Java being fairly new, the textbooks that we were assigned were absolutely terrible. Just garbage books.
Let’s also not forget efforts to make test scores less important.
If you actually tied outcomes to test scores all these “causes” would disappear.
Tied test scores to what outcomes? Bush the second tried tying test results to one kind of outcomes (school funding), and the result was schools getting worse.
Like in South Korea?
AI, screen time, short form text/video, normalization of recreational drugs, post-truth, there's no one cause but there's lots of things that are not helping
WolframAlpha was able to do advanced calculus in 2009 fully for free. LLMs shouldn’t have affected math performance at all because the tools were already readily available for a decade at that point.
totally false equivalency. wolfram alpha cannot one shot word problems for one, and there are many many more.
It got you 90% of the way there once you’d translated the problem. I can’t imagine LLMs ultimately saved more time per problem than WA did.
1 reply →
Wolfram is a tool, and LLM even without a harness is much closer to an agent.