This is the seventh in a series of AI in Education specials, where I speak to the world’s leading experts about how AI is changing teaching and learning.
I've found all of these articles very interesting, but I think this last one "brings it home" in terms of the future of AI being useful in the classroom. I too believe 'option 4' will be the winner, look forward to seeing the results.
Given that....: "the evidence that unconstrained chatbots hurt learning is now strong enough that I think any school allowing students to use ChatGPT for maths homework should be worried." ... How on Earth can a school actually enforce allowing use of AI at home or not? To me the conclusion is clear : unsupervised work has no value in an AI age. I'm more than happy for someone to disagree with me and justify that homework has value.
Thanks for replying and yes I read the article last year, which came out with a menu of four different choices. Since then AI has progressed even further, and I'm increasingly of the opinion that unsupervised work has no value in any subject, not just mathematics. I also entirely agree with the view that homework does have value as a means of independent prepping for assessment, but we need to be careful about interpreting poor assessment results as a proxy for no independent work or as a proxy for relatively lower ability or as a proxy for stuck misconceptions.
The question I'd most like to see answered is: does the same results on AI vs human moderated learning apply to older generations?
My suspect is that there might be a hidden bias: nowadays students could tend to trust more AI than teachers or tutors*, so if they knew "who" was helping them in their learning, the outcome of this study would assess both pedagogy and personal (students') preference, not pedagogy only.
*The reason I believe this is that AI doesn't judge, while tutors and humans do.
Are we sure that an even softer, praiseful and human-less learning will help our students grow as happy and confident beings?
Isn't it better for them to learn how to face the fear of judgement, to learn how to understand information from imperfect teachers, to trust them but always use their own critical thought 'cause a teacher is never perfect?
I fear the day education will become a productive process... and the more I delve into English pedagogy the more I understand why the anglo-led world is really giving credit to these technologies as possible "learning enhancers".
Other pedagogies, which aren't content-based, are already many steps ahead: try to ask any italian student who went abroad how was school/university, they'll always answer one thing: "easy".
This different perspective of classic italian pedagogy is based on some ideas and common practices: most of the skills human needs are hard to impossible to teach. A teacher should inspire and give chances to try, setting high expectations and letting the kids fix their own mistakes by the only method life uses: failure and reattempt. Eventually, everyone always learns alone (and that's a truth, not an idea: the moment you intake knowledge is personal and private).
We should make education life-like, because, ultimately, learning is about growing, not about learning.
In the attempt of helping students, I'd opt for this rather than <<optimize the test's results by enforcing new AI driven teaching>>, which looks much alike a bot-made string of characters...
What are your thoughts on the Teach Aid AI platform for schools? It claims to plan in one click, starting with a unit plan, moving to lessons and resources?
Not sure if that's an open question to all or directed to Craig.
FWIW though, and without having looked into it, sounds like an 'aggregator' to me: know the curriculum; divvy out into the appropriate number of units; collate existing, or generate new, resources to support the content of said units.
Bit like using AI to plan a holiday.
I could of course be doing it a massive disservice, but the planning part isn't the real challenge in today's maths classes, it's the comprehension, embedding and confidence in maths that's needed and I think the "Option 4" from the article is the way forward there to assist teachers based on what the AI 'knows' about the student to help them succeed.
I've found all of these articles very interesting, but I think this last one "brings it home" in terms of the future of AI being useful in the classroom. I too believe 'option 4' will be the winner, look forward to seeing the results.
Given that....: "the evidence that unconstrained chatbots hurt learning is now strong enough that I think any school allowing students to use ChatGPT for maths homework should be worried." ... How on Earth can a school actually enforce allowing use of AI at home or not? To me the conclusion is clear : unsupervised work has no value in an AI age. I'm more than happy for someone to disagree with me and justify that homework has value.
I 100% agree. I wrote about it (and a potential solution) here: https://eedi.substack.com/p/is-maths-homework-dead?r=1hg4l1&utm_medium=ios
Thanks for replying and yes I read the article last year, which came out with a menu of four different choices. Since then AI has progressed even further, and I'm increasingly of the opinion that unsupervised work has no value in any subject, not just mathematics. I also entirely agree with the view that homework does have value as a means of independent prepping for assessment, but we need to be careful about interpreting poor assessment results as a proxy for no independent work or as a proxy for relatively lower ability or as a proxy for stuck misconceptions.
The question I'd most like to see answered is: does the same results on AI vs human moderated learning apply to older generations?
My suspect is that there might be a hidden bias: nowadays students could tend to trust more AI than teachers or tutors*, so if they knew "who" was helping them in their learning, the outcome of this study would assess both pedagogy and personal (students') preference, not pedagogy only.
*The reason I believe this is that AI doesn't judge, while tutors and humans do.
Are we sure that an even softer, praiseful and human-less learning will help our students grow as happy and confident beings?
Isn't it better for them to learn how to face the fear of judgement, to learn how to understand information from imperfect teachers, to trust them but always use their own critical thought 'cause a teacher is never perfect?
I fear the day education will become a productive process... and the more I delve into English pedagogy the more I understand why the anglo-led world is really giving credit to these technologies as possible "learning enhancers".
Other pedagogies, which aren't content-based, are already many steps ahead: try to ask any italian student who went abroad how was school/university, they'll always answer one thing: "easy".
This different perspective of classic italian pedagogy is based on some ideas and common practices: most of the skills human needs are hard to impossible to teach. A teacher should inspire and give chances to try, setting high expectations and letting the kids fix their own mistakes by the only method life uses: failure and reattempt. Eventually, everyone always learns alone (and that's a truth, not an idea: the moment you intake knowledge is personal and private).
We should make education life-like, because, ultimately, learning is about growing, not about learning.
In the attempt of helping students, I'd opt for this rather than <<optimize the test's results by enforcing new AI driven teaching>>, which looks much alike a bot-made string of characters...
Thanks for reading,
Marco Arrighetti
What are your thoughts on the Teach Aid AI platform for schools? It claims to plan in one click, starting with a unit plan, moving to lessons and resources?
Not sure if that's an open question to all or directed to Craig.
FWIW though, and without having looked into it, sounds like an 'aggregator' to me: know the curriculum; divvy out into the appropriate number of units; collate existing, or generate new, resources to support the content of said units.
Bit like using AI to plan a holiday.
I could of course be doing it a massive disservice, but the planning part isn't the real challenge in today's maths classes, it's the comprehension, embedding and confidence in maths that's needed and I think the "Option 4" from the article is the way forward there to assist teachers based on what the AI 'knows' about the student to help them succeed.
Thanks, I 100% agree around where the challenges are