Does AI actually help students learn? A recent experiment in a high school provides a cautionary tale.
Researchers at the University of Pennsylvania found that Turkish high school students who had access to ChatGPT while doing practice math problems did worse on a math test compared with students who didn’t have access to ChatGPT. Those with ChatGPT solved 48 percent more of the practice problems correctly, but they ultimately scored 17 percent worse on a test of the topic that the students were learning.
A third group of students had access to a revised version of ChatGPT that functioned more like a tutor. This chatbot was programmed to provide hints without directly divulging the answer. The students who used it did spectacularly better on the practice problems, solving 127 percent more of them correctly compared with students who did their practice work without any high-tech aids. But on a test afterwards, these AI-tutored students did no better. Students who just did their practice problems the old fashioned way — on their own — matched their test scores.
Taking too many shortcuts doesn’t help anyone learn anything.
Kids who take shortcuts and don’t learn suck at recalling knowledge they never had…
I took German in high school and cheated by inventing my own runic script. I would draw elaborate fantasy/sci-fi drawings on the covers of my notebooks with the German verb declensions and whatnot written all over monoliths or knight’s armor or dueling spaceships, using my own script instead of regular characters, and then have these notebook sitting on my desk while taking the tests. I got 100% on every test and now the only German I can speak is the bullshit I remember Nightcrawler from the X-Men saying. Unglaublich!
I just wrote really small on a paper in my glasses case, or hidden data in the depths of my TI86.
We love Nightcrawler in this house.
Meanwhile the teacher was thinking, “interesting tactic you’ve got there, admiring your art in the middle of a test”
God knows what he would have done to me if he’d caught me. He once threw an eraser at my head for speaking German with a Texas accent. In his defense, he grew up in a post-war Yugoslavian concentration camp.
Must be same era, my elderly off the boat Italian teacher in 90s Brooklyn used to hit me with his cane.
Actually if you read the article ChatGPT is horrible at math a modified version where chatGPT was fed the correct answers with the problem didn’t make the kids stupider but it didn’t make them any better either because they mostly just asked it for the answers.
The only reason we’re trying to somehow compromise and allow or even incorporate cheating software into student education is because the tech-bros and singularity cultists have been hyping this technology like it’s the new, unstoppable force of nature that is going to wash over all things and bring about the new Golden Age of humanity as none of us have to work ever again.
Meanwhile, 80% of AI startups sink and something like 75% of the “new techs” like AI drive-thru orders and AI phone support go to call centers in India and Philippines. The only thing we seem to have gotten is the absolute rotting destruction of all content on the internet and children growing up thinking it’s normal to consume this watered-down, plagiarized, worthless content.
deleted by creator
Good tl;dr
Because AI and previously google searches are not a substitute for having knowledge and experience. You can learn by googling something and reading about how something works so you can figure out answers for yourself. But googling for answers will not teach you much. Even if it solves a problem, you won’t learn how. And won’t be able to fix something in the future without googling th answer again.
If you dont learn how to do something, you won’t be experienced enough to know when you are doing it wrong.
I use google to give me answers all the time when im problem solving. But i have to spend a lot more time after the fact to learn why what i did fixed the problem.
Nope, this doesn’t work like this. sometimes you need someone to explain, specially on math, youtube can take that spot, but not always.
That’s what i am saying. You need to learn it. If someone explains it to you, then you are learning. If someone gives you the answer, then you dont understand it, so you are less good at said something.
You agree with me…
Maybe, if the system taught more of HOW to think and not WHAT. Basically more critical thinking/deduction.
This same kinda topic came up back when I was in middle/highschool when search engines became wide spread.
However, LLM’s shouldn’t be trusted for factual anything, same as Joe blows blog on some random subject. Did they forget to teach cross referencing too? I’m sounding too bitter and old so I’ll stop.
However, LLM’s shouldn’t be trusted for factual anything, same as Joe blows blog on some random subject.
Podcasts are 100% reliable tho
ChatGPT lies which is kind of an issue in education.
As far as seeing the answer, I learned a significant amount of math by looking at the answer for a type of question and working backwards. That’s not the issue as long as you’re honestly trying to understand the process.
Would kids do better if the AI doesn’t hallucinate?
Paradoxically, they would probably do better if the AI hallucinated more. When you realize your tutor is capable of making mistakes, you can’t just blindly follow their process; you have to analyze and verify their work, which forces a more complete understanding of the concept, and some insight into what errors can occur and how they might affect outcomes.
Would snails be happier if it kept raining? What can we do to make it rain forever and all time?
Kids who use ChatGPT as a study assistant do worse on tests
But on a test afterwards, these AI-tutored students did no better. Students who just did their practice problems the old fashioned way — on their own — matched their test scores
Headline: People who flip coins have a much worse chance of calling it if they call heads!
Text: Studies show that people who call heads when flipping coins have an even chance of getting it right compared to people who do the old fashion way of calling tails.
You skipped the paragraph where they used two different versions of LLMs in the study. The first statement is regarding generic ChatGPT. The second statement is regarding an LLM designed to be a tutor without directly giving answers.
I didn’t skip it. If you are going to use a tool, use it right. “Study shows using the larger plastic end of screwdriver makes it harder to turn screws than just using fingers to twist them. Researchers caution against using screwdriver to turn screws.”
That’s not the fault of the user/students, though. They’re different tools. One is outright worse than not using it. Neither produce lasting benefits.
Headline: Screwdrivers better than hammers for screws.
Text: When craftspeople were trained using hammers with screwdriver bits duct-taped to them, they were able to perform the task, but were not able to keep pace with people using screwdrivers. Another team was given power drills, which were effective in practice. However, these did not produce any benefit once all people were given screwdrivers.
Thats a modified version, it says using unmodified ChatGPT results in 17% worse scores
While I get that, AI could be handy for some subjects, where you wont put your future on. However using it extinsively for everything is quite an exaggeration.
This isn’t a new issue. Wolfram alpha has been around for 15 years and can easily handle high school level math problems.
Only old farts still use Wolfram
What do young idiots use?
ChatGPT apparently lol
I can’t remember, but my dad said before he retired he would just pirate Wolfram because he was too old to bother learning whatever they were using. He spent 25 years in academia teaching graduate chem-e before moving to the private sector. He very briefly worked with one of the Wolfram founders at UIUC.
Edit: I’m thinking of Mathematica, he didn’t want to mess with learning python.
Where did you think you were?
Except wolfram alpha is able to correctly explain step by step solutions. Which was an aid in my education.
Shocked, I tell you!
TLDR: ChatGPT is terrible at math and most students just ask it the answer. Giving students the ability to ask something that doesn’t know math the answer makes them less capable. An enhanced chatBOT which was pre-fed with questions and correct answers didn’t screw up the learning process in the same fashion but also didn’t help them perform any better on the test because again they just asked it to spoon feed them the answer.
references
ChatGPT’s errors also may have been a contributing factor. The chatbot only answered the math problems correctly half of the time. Its arithmetic computations were wrong 8 percent of the time, but the bigger problem was that its step-by-step approach for how to solve a problem was wrong 42 percent of the time.
The tutoring version of ChatGPT was directly fed the correct solutions and these errors were minimized.
The researchers believe the problem is that students are using the chatbot as a “crutch.” When they analyzed the questions that students typed into ChatGPT, students often simply asked for the answer.
I don’t even know of this is ChatGPT’s fault. This would be the same outcome if someone just gave them the answers to a study packet. Yes, they’ll have the answers because someone (or something) gave it to them, but won’t know how to get that answer without teaching them. Surprise: For kids to learn, they need to be taught. Shocker.
Not limited to kids.
I’ve found chatGPT to be a great learning aid. You just don’t use it to jump straight to the answers, you use it to explore the gaps and edges of what you know or understand. Add context and details, not final answers.
The study shows that once you remove the LLM though, the benefit disappears. If you rely on an LLM to help break things down or add context and details, you don’t learn those skills on your own.
I used it to learn some coding, but without using it again, I couldn’t replicate my own code. It’s a struggle, but I don’t think using it as a teaching aid is a good idea yet, maybe ever.
I wouldn’t say this matches my experience. I’ve used LLMs to improve my understanding of a topic I’m already skilled in, and I’m just looking to understand something nuanced. Being able to interrogate on a very specific question that I can appreciate the answer to is really useful and definitely sticks with me beyond the chat.
There are lots of studies out there, and many of them contradict each other. Having a study with references contribute to the discussion, but it isn’t the absolute truth.
There’s a bunch of websites that give you the answers to most homework. You can just Google the question and find the answers pretty quickly. I assume the people using chatgtp to “study” are just cheating on homework anyway.
Did those using tutor AI spend less time on learning? That would have been worth measuring
Interesting thought, I would be curious about this too.
Yeh because it’s just like having their dumb parents do homework for them
