Hi Chris! I'm excited to find your substack and reengage with you here. My (small, overworked) English dept. has really been spinning its wheels to get a handle on AI, and your thoughts have been really helpful to me. I especially loved this post about centering and articulating the values of our field.
One thought that occurred to me as a value proposition to students had to do with the power of examples as a learning tool. LLMs are trained on, essentially, millions of examples, which is why they can reproduce the form of genre-specific writing, even if the content is a hodge-podge of clichés, platitudes, and generalizations. The reason GPT can write a better memo than some of my professional writing students is simply that it has digested many more examples of a memo than the average 18 or 19 year old college student. But humans also learn much more quickly and require far fewer examples than an LLM to learn a genre and internalize its expectations. One of the changes I'm going to make to my teaching, especially in a genre-focused class like professional writing, is to simply provide more good examples than I have in the past to make students feel more empowered and less likely to turn to an automated text generator.
Also--a quick book recommendation for everybody in the comments: I just started reading Literary Theory for Robots: How Computers Learned to Write, which was released earlier this year, and it's quite good and accessible.
I love this point about the speed of learning from examples, Adam. Everyone is in awe over LLMs' speed, but it seems to me we're much better at inferring from a few examples than LLMs, at least for now. (Always at least for now.) Thanks also for the book recommendation!
This is an inspiring (albeit occasionally scary) read. I am inspired, Chris, because your post gave me a better handle on how to engage my students (and fellow faculty) on the issue of AI (non) integration and use. I've been trying to get to the why question, and you helped me think though that better. I'm going to take it a step further and let students ask and answer that question in relation to why they're in college, why they want to learn to write (and every skill and knowledge it embodies). And I was a bit scared because I think there is a significant chance that academia and society will adopt this technology (like it did social media) without thinking through in ways you've been advocating for and showing us. But I guess all we have the best we can do. Thank you for another thought-provoking post.
Thanks, Shyam. I share your fear, and to me, the only way to counteract that unthinking adoption of the technology is to be really vocal about what we do want education/academia to be (as well as society at large) and then build networks of solidarity to continue advocating for those values.
Hi Chris! I'm excited to find your substack and reengage with you here. My (small, overworked) English dept. has really been spinning its wheels to get a handle on AI, and your thoughts have been really helpful to me. I especially loved this post about centering and articulating the values of our field.
One thought that occurred to me as a value proposition to students had to do with the power of examples as a learning tool. LLMs are trained on, essentially, millions of examples, which is why they can reproduce the form of genre-specific writing, even if the content is a hodge-podge of clichés, platitudes, and generalizations. The reason GPT can write a better memo than some of my professional writing students is simply that it has digested many more examples of a memo than the average 18 or 19 year old college student. But humans also learn much more quickly and require far fewer examples than an LLM to learn a genre and internalize its expectations. One of the changes I'm going to make to my teaching, especially in a genre-focused class like professional writing, is to simply provide more good examples than I have in the past to make students feel more empowered and less likely to turn to an automated text generator.
Also--a quick book recommendation for everybody in the comments: I just started reading Literary Theory for Robots: How Computers Learned to Write, which was released earlier this year, and it's quite good and accessible.
I love this point about the speed of learning from examples, Adam. Everyone is in awe over LLMs' speed, but it seems to me we're much better at inferring from a few examples than LLMs, at least for now. (Always at least for now.) Thanks also for the book recommendation!
This is an inspiring (albeit occasionally scary) read. I am inspired, Chris, because your post gave me a better handle on how to engage my students (and fellow faculty) on the issue of AI (non) integration and use. I've been trying to get to the why question, and you helped me think though that better. I'm going to take it a step further and let students ask and answer that question in relation to why they're in college, why they want to learn to write (and every skill and knowledge it embodies). And I was a bit scared because I think there is a significant chance that academia and society will adopt this technology (like it did social media) without thinking through in ways you've been advocating for and showing us. But I guess all we have the best we can do. Thank you for another thought-provoking post.
Thanks, Shyam. I share your fear, and to me, the only way to counteract that unthinking adoption of the technology is to be really vocal about what we do want education/academia to be (as well as society at large) and then build networks of solidarity to continue advocating for those values.
Great stuff, Chris. You are really emerging as a thought leader in this area.
Thanks, Paul! Just trying to help as best as I can