I think this is a fab way to use AI, I'm all in favour of usage cases where we are encouraged to think more rather than subbing out our brains.
I think some things to consider about the tutorial system (and therefore how you might refine your Oxford-tutor AI plan)
-You very rarely have a tutorial on one book. Actually I can't remember ever doing that.
Each week is either a theme (with several primary texts and scholarship) or one main text and supplementary scholarship. This is absolutely not to say that you shouldn't dig deeper into individual texts in this way, but rather to illustrate the practice of taking in multiple perspectives and orienting yourself in the landscape- what is the context, what do academics think of the work? what are the debates surrounding it? Maybe after a deep dive you could ask for a list of secondary sources on the text or a similar/contrasting primary text to compare it to.
-Before most tutorials you do a piece of work on your own. Usually an essay, that synthesises everything you have read. Usually it's on a question or a statement, sometimes its more 'write what interests you'
Perhaps getting AI to give you an interesting or thought provoking question based on your new reading list and doing a short essay is an option.
This act of creation without outside input (other than the reading itself) seems to me fundamental to the system. In much the same way as you learn better by doing, creating something from your reading will improve understanding and retention.
And, more importantly, if you rush into being 'tutored' by anyone, human or AI, you are giving away something precious, your first impression. Your perspective.
Which brings me on to my final point...
-Your opinion matters.
I noticed- although to be fair its a small sample in this article- that the questions are very much more about what the scholars or philosophers say. Which while very important seems to be to be only the beginning of an Oxford tutorial. Make sure you move on to what you think, how you would define an intellectual etc., do you agree with the original author? Do you like their ideas but think their writing is as dense You may feel unqualified to have an opinion, but try to form one anyway- you can always change your mind (sometimes with the persuasion of your tutor..).
Sometimes what you have written or thought will be entirely misguided- which AI will be able to tell you when you submit your essay to it- but if I'm honest, the weeks where I turned in an essay that entirely missed the point, where the weeks when I learnt the most. Will never forget Xenophon's constitution of the Spartans, thanks to my car crash of an essay haha!
I sense in a lot of autodidacts a real fear of being wrong, but if I learned anything from my Oxford education it was to chill out, have the courage to be honest and say when you don't know or understand something fully, have the curiosity to play with knowledge and opinions, and never take what another scholar says for - heck that does sound a lot like the intellectual when I come to type it out...
This is such a helpful comment, Molly. Chill out, indeed! We all need to hear that.
Thank you for further clarifying the tutorial system for me. To your point about the importance of forming and articulating your own opinion, I wholeheartedly agree.
In my experiments, I have been asked questions about what I think of the material, especially when I used Chat-GPT to converse with me about Great Expectations. I bet I could make the AI tutor include these types of questions for non-fiction book testing as well. I'll report back in my next piece about this.
Have you considered using a GPT or Project or just attaching sections of the book to the chat? Or have you used NotebokLM? I wrote about this technique a few months ago.
Bob - glad you found it helpful. I’ve been deep in the AI weeds since ChatGPt came to the forefront and try to keep on top of everything and it’s STILL overwhelming. The speed with which this tech is taking off is fairly unprecedented. Don’t abandon all hope - paper and pencil skills won’t disappear. But I think your instincts to see what all the fuss is about is a good one. At its best, AI is a remarkable collaborator. It’s a shame (though not surprising) that the cheating conversation dominates because it’s so much more than that despite what the critics say.
Steve, thanks for sharing your article, it dovetails nicely with Sam's post.
Like many of us, we're dabbling with AI and learning on the fly. My first few attempts at messing with NotebookLM absolutely knocked my socks off - It's ability to build a conversational 2-person podcast formulated on the specific content it was fed, along with a it building a study guide, FAQ's and a mock quiz, for a complete AI novice like me, was remarkable.
As much as I'd like to keep my head in the sand and continue reading, taking notes with pencil & paper and synthesizing information like I've done to this point in my life, I know that I have to get past being a prude and recognize that I need to find responsible, effective ways of harnessing the technology for good.
Sam & Steve, Thank You both for sharing your work - Cheers!
Of course— and thank you, Stephen and Bob, for these thoughtful additions to the conversation!
Stephen, your article is fascinating. It seems we both want to help people use AI tools to interact more deeply with the content. Love how some students were able to go deeper into the subtopics that interested them most. Seems like another valuable use case: facilitating self-directed deep dives that aren't usually encouraged in a traditional classroom setting. Your students are lucky to have you!
Bob, that 2-person podcast feature sounds so useful. I might have to try it out as a way to support my reading. Hit the brain from all angles, right? A part of me wishes Chat-GPT didn't exist, but you're right, we can't hide from it. Have to harness its potential for good.
Wow — that's one of the nicest compliments I've received! Thank you. You're so right. I often learn so much from the comments section. Nothing's more helpful for refining my ideas than the smart, critical thinkers who are nice enough to engage with my work.
I wish Oxford had shared this guide with their alumni too - but thanks for doing the communication for them 😁 I was fortunate to experience the tutoring system in real life. Quite often it’d be not a teaching assistant but a senior lecturer or even professor, so it was also terrifying at times (especially in a non native language). At least I assume the AI will always be less intimidating!
By the way, if you were to use the memory function on ChatGPT, you could probably run a second round of these conversations about the previous one - say a month later, to test progress? I’ll need to play around with it more.
That's a great idea—I'll have to experiment with the memory function. And yes, AI is certainly less intimidating than any oral exam 🙃 I hate public speaking, so I've always admired those who can get in front of people and talk about difficult subjects, especially when those experts who are grading you.
Great idea. Like you, my undergrad days are long gone, but I constantly worry I don’t get as much out of my reading as I’d like to. I recently asked ChatGPT to create a semester-long undergrad course on gothic literature that I could follow by myself. However, I was concerned about not having a teacher to correct/guide/challenge me. I’ll go back to it now and ask ChatGPT to be that teacher.
That's a great use case! The feeling is real. Hopefully, AI's ability to democratize noble features of higher education offsets its nasty habit of turning some of us into semi-lobotomized drones.
Love that — I too am working on some literary writing projects, and I don't intend on ever letting it write for me (that'd deeply depress me). I also still like to re-read books and transcribe my marginalia and underlines into a physical notebook. It's all about figuring out how the tool works in our own unique self-study workflows. So far, this ability to generate and grade tests for me is probably as far as I'm willing to let it into my humanities self-education.
I think this is a fab way to use AI, I'm all in favour of usage cases where we are encouraged to think more rather than subbing out our brains.
I think some things to consider about the tutorial system (and therefore how you might refine your Oxford-tutor AI plan)
-You very rarely have a tutorial on one book. Actually I can't remember ever doing that.
Each week is either a theme (with several primary texts and scholarship) or one main text and supplementary scholarship. This is absolutely not to say that you shouldn't dig deeper into individual texts in this way, but rather to illustrate the practice of taking in multiple perspectives and orienting yourself in the landscape- what is the context, what do academics think of the work? what are the debates surrounding it? Maybe after a deep dive you could ask for a list of secondary sources on the text or a similar/contrasting primary text to compare it to.
-Before most tutorials you do a piece of work on your own. Usually an essay, that synthesises everything you have read. Usually it's on a question or a statement, sometimes its more 'write what interests you'
Perhaps getting AI to give you an interesting or thought provoking question based on your new reading list and doing a short essay is an option.
This act of creation without outside input (other than the reading itself) seems to me fundamental to the system. In much the same way as you learn better by doing, creating something from your reading will improve understanding and retention.
And, more importantly, if you rush into being 'tutored' by anyone, human or AI, you are giving away something precious, your first impression. Your perspective.
Which brings me on to my final point...
-Your opinion matters.
I noticed- although to be fair its a small sample in this article- that the questions are very much more about what the scholars or philosophers say. Which while very important seems to be to be only the beginning of an Oxford tutorial. Make sure you move on to what you think, how you would define an intellectual etc., do you agree with the original author? Do you like their ideas but think their writing is as dense You may feel unqualified to have an opinion, but try to form one anyway- you can always change your mind (sometimes with the persuasion of your tutor..).
Sometimes what you have written or thought will be entirely misguided- which AI will be able to tell you when you submit your essay to it- but if I'm honest, the weeks where I turned in an essay that entirely missed the point, where the weeks when I learnt the most. Will never forget Xenophon's constitution of the Spartans, thanks to my car crash of an essay haha!
I sense in a lot of autodidacts a real fear of being wrong, but if I learned anything from my Oxford education it was to chill out, have the courage to be honest and say when you don't know or understand something fully, have the curiosity to play with knowledge and opinions, and never take what another scholar says for - heck that does sound a lot like the intellectual when I come to type it out...
*never take what another scholar says for gospel
This is such a helpful comment, Molly. Chill out, indeed! We all need to hear that.
Thank you for further clarifying the tutorial system for me. To your point about the importance of forming and articulating your own opinion, I wholeheartedly agree.
In my experiments, I have been asked questions about what I think of the material, especially when I used Chat-GPT to converse with me about Great Expectations. I bet I could make the AI tutor include these types of questions for non-fiction book testing as well. I'll report back in my next piece about this.
Have you considered using a GPT or Project or just attaching sections of the book to the chat? Or have you used NotebokLM? I wrote about this technique a few months ago.
https://fitzyhistory.substack.com/p/my-experiment-with-guided-reading
Bob - glad you found it helpful. I’ve been deep in the AI weeds since ChatGPt came to the forefront and try to keep on top of everything and it’s STILL overwhelming. The speed with which this tech is taking off is fairly unprecedented. Don’t abandon all hope - paper and pencil skills won’t disappear. But I think your instincts to see what all the fuss is about is a good one. At its best, AI is a remarkable collaborator. It’s a shame (though not surprising) that the cheating conversation dominates because it’s so much more than that despite what the critics say.
Steve, thanks for sharing your article, it dovetails nicely with Sam's post.
Like many of us, we're dabbling with AI and learning on the fly. My first few attempts at messing with NotebookLM absolutely knocked my socks off - It's ability to build a conversational 2-person podcast formulated on the specific content it was fed, along with a it building a study guide, FAQ's and a mock quiz, for a complete AI novice like me, was remarkable.
As much as I'd like to keep my head in the sand and continue reading, taking notes with pencil & paper and synthesizing information like I've done to this point in my life, I know that I have to get past being a prude and recognize that I need to find responsible, effective ways of harnessing the technology for good.
Sam & Steve, Thank You both for sharing your work - Cheers!
Of course— and thank you, Stephen and Bob, for these thoughtful additions to the conversation!
Stephen, your article is fascinating. It seems we both want to help people use AI tools to interact more deeply with the content. Love how some students were able to go deeper into the subtopics that interested them most. Seems like another valuable use case: facilitating self-directed deep dives that aren't usually encouraged in a traditional classroom setting. Your students are lucky to have you!
Bob, that 2-person podcast feature sounds so useful. I might have to try it out as a way to support my reading. Hit the brain from all angles, right? A part of me wishes Chat-GPT didn't exist, but you're right, we can't hide from it. Have to harness its potential for good.
Fantastic. I can't decide which is better, your article or the comments from your readers. You've got some really smart readers!
Wow — that's one of the nicest compliments I've received! Thank you. You're so right. I often learn so much from the comments section. Nothing's more helpful for refining my ideas than the smart, critical thinkers who are nice enough to engage with my work.
I wish Oxford had shared this guide with their alumni too - but thanks for doing the communication for them 😁 I was fortunate to experience the tutoring system in real life. Quite often it’d be not a teaching assistant but a senior lecturer or even professor, so it was also terrifying at times (especially in a non native language). At least I assume the AI will always be less intimidating!
By the way, if you were to use the memory function on ChatGPT, you could probably run a second round of these conversations about the previous one - say a month later, to test progress? I’ll need to play around with it more.
That's a great idea—I'll have to experiment with the memory function. And yes, AI is certainly less intimidating than any oral exam 🙃 I hate public speaking, so I've always admired those who can get in front of people and talk about difficult subjects, especially when those experts who are grading you.
It adds a lot of pressure on students by design. 15 years later I’m glad I don’t need that anymore to keep on learning!
This is pretty much how I use AI, though not as structured.
Insightful and excellent nuts and bolts on how to… thanks for doing the investigative legwork for us!
Thank you :)
Thank you for this. I'm currently using Chat GPT to help me write a novel, and I will now start using it to teach me about the books i'm reading.
This sounds promising. I’m looking forward to your next post.
I think so too!
Great idea. Like you, my undergrad days are long gone, but I constantly worry I don’t get as much out of my reading as I’d like to. I recently asked ChatGPT to create a semester-long undergrad course on gothic literature that I could follow by myself. However, I was concerned about not having a teacher to correct/guide/challenge me. I’ll go back to it now and ask ChatGPT to be that teacher.
That's a great use case! The feeling is real. Hopefully, AI's ability to democratize noble features of higher education offsets its nasty habit of turning some of us into semi-lobotomized drones.
Love that — I too am working on some literary writing projects, and I don't intend on ever letting it write for me (that'd deeply depress me). I also still like to re-read books and transcribe my marginalia and underlines into a physical notebook. It's all about figuring out how the tool works in our own unique self-study workflows. So far, this ability to generate and grade tests for me is probably as far as I'm willing to let it into my humanities self-education.