To supercharge learning, look to play
nautil.us
nautil.us
My wife is an educator (public 1st grade teacher in California). We just had the discussion that next year will likely be her last year in that profession. The stress is too high, support too low, kid's behavior issues are skyrocketing, parents are getting downright violent/threatening to the staff...it's just a fuckin mess. Many of her colleagues are leaving the profession. I expect there to be a growing deficit of teachers in the coming years.
We're hoping we can get her working on education materials/maybe tutoring as a side business during her sabbatical (maybe retirement)?
I bring this up because what I've been thinking about lately, is that with the recent explosion of Large Language Models and their inevitable rapid evolution, that to me it's pretty clear education needs to go down the path of AI-based, automatically customized and tuned and guided, computer based education for primary students. We're right around the corner from AI automatically generating highly interactive learning courses for children, that will fundamentally reshape the notion of classroom-based education.
The path to adoption will probably be a mess because of the bureaucracy in education in general, but maybe that means more people will explore private/home-based education paths combined with outlets for social interaction for their kids (maybe there will be a boom in youth sports?)
Random Saturday morning daydreaming here, curious what others think.
But below that level, the topics that are being taught are very standardized and have been studied to death, and there already exists a ton of interactive learning tools, toys, projects, and games. Teachers below middle-school level are mostly babysitting and trying to just get kids to learn the rules and pay attention.
I am concerned for the education of the next generation of kids. Stories like yours are all too common.
Can you explain your process for using it? Is it done in a session or do you ask random questions throughout the day? Specific topics or general?
when I study or learn stuff, I have questions. I'm naturally curious and desire to fully understand something, but also a topic may encourage me to explore strange other avenues or branches or topics until I'm satisfied and can return.
Reading from books or videos can be frustrating for obvious reasons. When tutored, I need an instructor who is not bound to "lesson one page one, now lesson one page two" blind and sequential approach.
Chatgpt has been nothing less than a God send.
E.g. I've tried to learn French 5 times over 25 years living in Canada. Always failed. I tried Michel Thomas self study, Berlitz group classroom, duolingo, even private tutor. They were serious efforts, but I hated the language and never felt I am even close to understanding it.
Now whenever I have the remotest slightest question or curiosity I ask chatgpt. And then move on with the sequential class satisfied. It's both emotionally satisfying and encouraging, and it allows me to have better understanding of each topic. It's been a genuine force multiplier.
Same with music theory (0.1% of piano teachers I've met are remotely capable or willing of answering questions "why". Two of them swore that "music theory" == "learning notation". Etc.) and, though far less, Python.
All the examples of hallucination I see on Internet are... Either controversial or pushy or edgy.
When asked a simple question on factual well-covered, non-controversial items, success rate seems very high.
Do we have a feel for safe and unsafe areas? Are there type of questions or ways of asking them that will produce high confidence?
(Maybe we have different ideas of "well covered". I am guessing if you ask it about the content of middle- or high-school textbooks it will generally be about as reliable as a typical middle- or high-school textbook. But that's the stuff I don't need an AI research assistant to find out about; I can just look those up in a search engine or wikipedia and quickly get an acceptable answer.)
I think it would be extremely dangerous to the general factual accuracy of claims found on the internet to have this widely adopted as a general learning tool, in its current form. Sort of "citogenesis on steroids". https://xkcd.com/978/
I ask it simple, well covered basic questions like list the types of minor chords, explain why melodic minor is different ascending and descending, what are the types of chords, explain seventh chords, etc.
Please deconstruct m'apelle and s'apelle and tell me what they're made of with examples. Explain difference between Tu, te, and toi,with examples.when do you reverse pronoun and verb in French. What are the ways you can formulate a question in French. Please list reflexive pronouns in French with examples. What is the difference between secularism and humanism.
Yes, I can find all these on google or Wikipedia - that's for me the point. I don't ask it any questions that wouldn't be solidly covered in its training set or would have controversy or judgment,or would be obscure my covered in a single scientific paper etc. But it's the seamless and fast method of obtaining meaningful and succinct answers to my specific questions that makes it completely different.
With google, if I have a question, I am now in googling mode, avoiding first three sponsored results and avoiding next six results that are SEO optimized stories of how they're learning French due to their first girlfriend's mother etc, and then sifting through pages to get a remotely applicable answer. It completely disrupts the learning. With chatgpt, it's a seamless integrated part of my learning.
>> what is laguerre geometry?
> Laguerre geometry is a branch of geometry developed by the French mathematician Edmond Laguerre in the 19th century. It is a non-Euclidean geometry that deals with the study of conic sections, curves, and surfaces in 3-dimensional space. Unlike Euclidean geometry, Laguerre geometry allows for the existence of lines and planes with complex, imaginary or dual coordinates. The Laguerre distance between two points is defined as the square root of the cross ratio of four points, and the geometry itself is often referred to as the geometry of cross-ratio invariants. The Laguerre geometry has applications in projective geometry, algebraic geometry, and physics.
The nontrivial parts of this are just wrong. Laguerre geometry is geometry where oriented lines are the basic objects rather than points, and other objects are constructed as envelopes of lines rather than loci of points. It is a kind of dual to inversive geometry (with "laguerre transformations" analogous to "möbius transformations").
If you don't mind, since I don't know the concept myself, where would you scale it on the axis of obscurity and controversy?
My own meagre check: In terms of obscurity - there's no Wikipedia entry for it, and a quick google search does not provide a super quick and simple definition to my ignorant and superficial check. There's tons on Laguerre transform but not on "Laguerre geometry". What there is, gives definitions that are not the same as the one you provided and is not consistently worded. I am not capable of reconciling and judging are they comparable or not, and are there more than one way of defining interpreting or explaining, nor exactly how wrong the AI definition is. At the same time, search for lobochevski geometry, which the most obscure one I know in my limited mathematical awareness, is amply and consistently covered - yet, still beyond the threshold of obscurity and consistency that I would use for AI. And search for Euclid geometry, which is what I'd expect AI to define correctly, is of course plentiful (which is why I'd expect AI to succeed, a bit circular there:).
So it may be simply that our expectations and need/use of LLM AI are different? To your original point, I'm indeed using it for stuff that's basic and common and it seems to work well there. If that's all it works well for, it's bloody amazing and very useful for many - even if definitely not (yet?)useful for all people, domains, or cases.
(As commonly happens, I am currently a bit stuck on making a bunch of diagrams...)
This is not controversial, but it is moderately obscure. (But not that obscure.)
* * *
I don’t have any particular “expectation”. I am just disappointed that the current versions of these tools do not give any indication about their level of certainty. I have asked a wide range of questions that prompted responses which were substantially bullshit. But if I hadn’t followed up might have seemed plausible enough.
Stuff like asking for translations of material written in Latin, asking about word etymologies, asking about court cases, asking for summaries of famous old books, ...
I have had a few conversations like:
“Who discovered thing A”
“It was X, who ....”
“wasn’t it Y?”
“Yes I am sorry. I was incorrect. It was Y, who ...”
“Wasn’t it actually Z?”
“Yes I am sorry. I was incorrect. It was Z, who ...”
If you try you can get it to make up nonsense about a whole string of false discoverers.
It would be better if it instead say “I have no idea about the precise answer, but people P and Q were working on similar topics around that time” or whatever.
To me, this implies that it may struggle when trying to make contextual connections. I.e. it may be better at answering “how” or “what” than “why it matters” or making value judgments
To my knowledge, the ChiSox have never had any Native American based logo and I suspect it is conflating them with another team.
Apparently an italian journalist asked about a writer and chatgpt said it was a mafia boss.
A teacher, real or virtual, needs to be right most of the times, and LLMs already provide that.
“The movement that wants to pause AI development should just tell school boards and DEI activists that AI will allow every child to learn at their own pace. Probably no faster way to put a total ban in motion than an appeal to toxic ‘anti-racism’ activism.”
It beats a broken human system that barely works at some point, the problem is we unempowered the teachers, let's empower the machines to teach then
Most families are doing their best, but being human is just hard sometimes.
I’m thinking more of 4–7-year-olds hitting each-other with sticks they pick up at the park, having temper meltdowns at trivial frustrations, refusing to follow instructions to stop doing obviously unsafe things, etc.
What do you think should be done in this sort of case? How does it get to this point, and what could be done to help kids like this before they reach the point of literally assaulting their teachers or other criminal-level violence? Most kinds of school punishments (detention, extra homework, suspension, ...) seem unlikely to really solve whatever issues this kid has, but teachers don’t have the extra bandwidth to be full-time social-worker caretakers of each specific kid.
With the foreword that I know this is a huge thorny mess of a problem with no easy solutions, there are three things that would make an immediate improvement.
#1, by far: Administrations as they currently are live in existential fear of a lawsuit that will destroy the entire district. Teachers are thrown under the bus with almost rabid fervor, as they are seen as fungible. This must stop, and real consequences need to be consistently enforced across the board. The genuinely unstable and violent need to be removed from the general school population. I do not know what to best do with them, and I will not even hazard a guess in this post. But I do know that they are holding the entirety of the education of the remaining 90% of the student body hostage to their whims, and that has to stop.
#2: The student teacher ratio needs to drop from 30+:1 to below 15:1, ideally even lower with the addition of aides along with regular teaching staff. The money is there- if you look at the per capita spend, the US is very high, even compared to other Western nations. We just blow it on literally anything other than paying teachers.
#3: Free, school-supplied breakfasts and lunches for the entire student body, no questions asked. Hunger is a huge deal, and hungry kids can't learn. Food instability affects somewhere between 20-50% of the students in the US, and it is a phenomenal return on outcome vs. $ spent.
Getting class sizes to 10:1 or 5:1 is quite the jobs program, and will lead to amazing long term outcomes for the kids.
Being a teacher is like being a nurse: it really takes the right kind of person to do the job well. Having more bodies in classrooms isn't a solution by itself.
One can also take a longer view and give better incentives to get the teaching credentials in the first place. It's not at all impossible to grow the number of teachers ethically.
----
No one here said schools have holes. The holes are in the govt which owns and controls public schools. Do you really think that is not relevant?
By your logic, schools don't tax people so it is irrelevant discussing taxation when it comes to improving schools.
We finally sat down and I helped her write a specifically worded email (finally got a good use out of all of that HR training as a manager! ha) to trigger the right response from her admin.
Any shitty single kid can ruin a classroom for the rest of their peers. And the system doesn't allow for any consequences. It's massively disappointing.
I have now dealt with dozens of public school teachers and while some were good, most were ambivalent and there were more garbage teachers than good ones.
The refrain of It's Not The Teachers rings hollow. It is the teachers. It may also be the administration, the parents, the community, the politicians, everybody, but it is still the teachers.
Also, people here should stop generalizing their experience in California to the country at large. MA and NJ have pretty good public schools, some of which are in dense and diverse regions of each state.
What everyone is talking around is the fact that the solution to "problem children" in the classroom, if you're removing them, is that you have to spend more money and more time working with them to get them off the "problem" track and re-integrated.
Because that's the stakes: that kid who disrupts the classroom, if you don't succeed, is your future welfare dependency, local crime problem, or incarceration cost and a huge amount of lost tax revenue.
Separation helps the problem kids as well. I have personally observed a high success rate in turning around kids once they get the more focused attention possible in a small class.
To be concrete, if you had 2 teachers and 60 kids with 5 of them "problematic", the most effective approach would be to divide them so you had the 55 cooperative learners in one room and the others in their own class.
Hence, nothing effective is allowed.
It's the same concept by which Google Maps will invariably tell you, "you are in the fastest path to for destination", even if you just took the dumbest wrong turn possible.
I am a former California educator switched to software engineer. The reasons I switched careers is directly because education is underfunded and under supported by school administration and I didn't see a sustainable future for myself there. Many other brilliant teachers I knew have also left or are leaving because it isn't worth the effort and stress for bare minimum pay. Teachers aren't leaving because of the students, they're leaving because of the systems. You're severely overworked and underpaid, with no backup or support from the institution, in overly large class sizes, and have poor curriculums that teach to outdated standardized tests that you have to stick to so the school can try to bargain for more funding.
I think a good system of well funded schools and teachers can get almost any student to a good place if they're enrolled at an early enough age and properly educated. Unfortunately there are increasingly common situations where a student has been let down by the school system for years and ends up falling too far behind their grade level to catch back up, but that is a failing of the school system and not their race or home life. I know because I have first hand experience in those elementary or middle schools, they just aren't able to serve students well with what they have (low wages lead to inexperienced or uneducated teaching staff, overfilled classrooms leads to not enough individual attention, poor learning environments, etc).
The school can work on behavioral issues, but then they go away on break, and come back with big regressions.
We should stop blaming everything on the schools. There are ways to raise a kid that are conducive to behaving well in that environment, and there are ways that are not conducive to that.
But we can’t blame it all on the parents either, they seem to have a lot less slack than I think US parents have had historically.
As for myself, I spent freshmen year stealing anything not bolted down and a few things that were. I was just bored and angry that the system wasn't treating me like it respected my potential. That was childish of me, but then again, I was a child.
A therapist recommended putting me in some advanced classes, and the behavior problems went away.
For these reasons I think that the adaptability of an AI generated curriculum might solve some of those behavior problems.
Then there are the other problems. Like maybe the student needs glasses to see the board or feels unsafe for some reason or is too busy working to feed their siblings. For those, we're going to need humans in the loop and compassion in the policies.
Modern educational curriculums have very little content for: critical thinking, logic, creativity, self-teaching, responsibility, civic responsibility, curiosity, compassion, self respect, and finding purpose. Instead of teaching how to use history to prevent making the same mistakes we make them memorize dates. Instead of teaching them to use writing as an emotional regulator and self therapy we teach them to read a paragraph quickly to answer a multiple choice question about it. Since standardized tests are easiest and cheapest to grade through a scantron, state tests emphasize facts and memorization.
The entire lecture and classroom school system 99% of us know teaches dependence over independence and blind obedience over critical thinking.
Sports are taught through a hyper-competitive lens, getting them ready for a possible professional sports career.
See the famous New York state teacher of the year John Gatto for more https://www.cantrip.org/gatto.html
I don't see ever see the wide adoption of an AI homeschooling solution coming to fruition due to this fact.
While many families did jump on homeschooling because of inexpensive computers, online courses and tutors, government schools jumped on online courses, assignments, instructors, etc. even when there is a teacher and two dozen classmates in the room. Government schools are far less pro-social, IME.
To whatever extent AI drives down the cost, it will accelerate the movement toward homeschooling, but I don't see it reducing the prosocial aspects of homeschooling, while it likely will in government schools.
When you say "empirically" though, I'm assuming there some info out there that backs this up? Would love to take a look, as homeschooling is on the table for us.
To your daydreaming:
I believe there will be a roll for AI tools in education, but they will not be replacements for in-person instruction. They will assist teachers in being better educators.
Unlike many professions where workers operate in concert with digital tools to unlock value that would be otherwise impossible to achieve, classrooms are essentially analog. The toolset I would like to build would start by mediating instruction with an assessment generation/proctoring tool that builds assessments in real time based on the lesson’s instructional content & conversational cues without significant curation required by the instructor (i.e. the system listens to the lesson as it unfolds and continuously generates/refines appropriate assessment content; the instructor should be able to complete a final curation and distribution of the assessment to students within a typical one to three minute instructional task/conversational break/etc).
Right now, we spend billions of dollars on standardized testing (with all the inherent issues of that high stakes, non-longitudinal format), and while the same content is being regularly assessed by classroom instructors, we broadly don’t accept those results as trustworthy. Such a system as I’m considering could inject that trust into classroom assessments, allow for more frequent/high quality/low effort assessments - and provide a beachhead for building additional purpose built assistive tools for instructors.
AI would likely provide utility to this tool, but the functional value cannot be separated from the fluid, branching path conversations of high quality in-person instruction.
Do you mind expanding on why that's clear? At even a first pass by ignoring the "AI" part and focusing on the computer-based approach, it is known and has been shown that screens are detrimental to cognitive development. As in detrimental to brain development and processes, not just learning.
> We're right around the corner from AI automatically generating highly interactive learning courses for children, that will fundamentally reshape the notion of classroom-based education.
How is it going to do this any better than humans? Part of the existing problem is that teachers can teach incorrect information. (I know your wife is a teacher. My mom was a teacher.) With auto-generated learning, what will be the check that the information is correct, following guidelines, not introducing biases and -isms, and not breaking laws and regulations? And I see this approach as doubling down on treating kids as automatons rather than teaching them critical and creative thinking.
The problems with education are multifaceted. Teachers are underpaid and overly managed with managed and standardized curriculums. Families do not have the proper support structures. Thus, schools are babysitters and sources of food for many. Social media is amplifying the worst parts of humanity and eating society from the inside out. We have an extremist/terrorist party insistent on not properly finding education and social support structures and destroying human rights.
It's unfathomable to me how machine learning based, auto-generated content somehow "solves" education without addressing these problems.
I think any parent should immediately withdraw their kid from a school that adopts these "AI". We can already see failures of forced technology in schools, such as PowerPoint and Chromebooks in schools. The myopic forcing of teachers to use PowerPoint was one of the many things that made my mom retire early from teaching.
There are many countries out there (Finland, France) that don't have this crisis in education and are not grasping for radical solutions. They are just incrementally improving the systems in place.
Why would the US need a completely different solution?
One reason is that the US is substantially more ethnically and culturally heterogenous than Finland or France. This means that a one-size-fits-all approach is much more likely to work over there than in the US.
No it isn't… but USA needs to be exceptional so you believe it is :)
It's pretty outlandish to say that claims that America is more diverse than that are made solely in pursuit of some exceptionalism.
Compare this with USA where there is 1 language.
Conversely, the most common USA racial group is non-Hispanic white (which groups many common languages/cultures, the most frequent of which are German/English/Polish/French). This group which is composed of all those different cultures comprises 60% of the US population.
> Compare this with USA where there is 1 language.
Nearly a quarter of American households speak a language besides English at home.
It's fine to say "there are things we can learn even though we are not as culturally homogeneous". It's fine to argue "cultural homogeneity isn't as big a deal for these things". But it's simply inconsistent with reality to claim that the USA is less ethnically and culturally diverse than Finland.
I teach 3rd, 4th, and 5th grade kids to code. Mostly code.org does the work of introducing code through play but the value of the teacher is in classroom management. Specific to programming, the value I bring is debugging experience and empathy. Instead of saying “you fix your problem like this” I focus on trying to understand/empathize with why the problem was made to begin with and then using that info to guide them to the correct fix.
Animals, including humans, instinctively play to learn, both in free play and guided play. They work particularly well when the kid is young. I don’t find “play” to have a negative connotation (in contrast to “work”). My hobbies as an adult often start with play.
However, there is something to be said about discipline. That also has some mixed connotations, so I will be clear. I am talking about discipline to mean the various inner psychology to focus, and sharpen one’s skills even through adversity. It includes what Angela Duckworth would call “grit”. Instilling this kind of discipline is not something I’d do at an early age, because it requires a sufficient level of mindfulness.
Discipline is how one can become truly great … but it is play that allows for a kind of creativity that allows one to generalize from a solid foundation. You need both to attain mastery.
As far as discipline goes, the value is being able to generate it from within. It’s a long grind, and it is not always fun.
I think the conflation with "punishment" has made it very difficult to teach self-discipline to kids and young adults.
Related things I learned from the martial art world are, “don’t cheat yourself”, and the idea of gongfu — mastery accumulated over time. And to bring some perspective to this, I also learned from the martial art world that “play” is crucial to develop the creative aspect of mastery. Some teachers will tell students to “play” with an idea, instead of running through a drill.
One of the best example of playing through learning for my kid was him learning the multiplication tables playing Roblox[1]
[1] https://www.roblox.com/games/10489578087/Multiplication-Game
If we look at children’s play until 100 years ago it’s all learning and prep for life in different forms - forming social, motor, observation, improvisational etc skills.
The gaming industry has hijacked this for its profit loop but ultimately games are about learning new skills and excelling at them.
Force feeding anything makes it super unfun.
One of my completed achievements was a sort of "typing tutor" game. I suppose I modeled it upon the game on my Casio calculator watch, and Missile Command. In my game, letters would fall from the sky, and pressing the correct key would destroy the letter and save Earth.
My father belittled it because it was a game and so, it couldn't be serious learning. Well Dad, I seriously learned some BASIC in order to get to a finished product and do a literal tape-out.
Raspberry Pi entered the chatNot only is this not a quote from Plato's writings, it doesn't even bear a passing resemblance to anything from Plato.
https://www.popatplay.org/post/launching-a-pedagogy-of-play-...
Spoiler: He thinks Zachtronics games (Spacechem, Shenzio IO, Opus Magnum etc) do this very well. The point is they're not explicitly educational games about specific topics, but they teach problem-solving skills in an engaging way.
I don’t have ADHD but (for different reasons) I struggle with similar issues. I also run (free) coaching/ranting sessions for people with similar problems.
That was actually Winston Churchill, not Plato: https://quoteinvestigator.com/2015/07/30/hour-play/
https://youtube.com/watch?v=gn-85vl-t5s
Is it a self-referential example of play?