September 15, 2023
Thanks to Alissa Simon, HMU Tutor, for today’s post.
I recently participated in a three day online festival hosted by The Chronicle of Higher Education. The topics to be discussed mentioned AI and technology which happen to be recent fascinations of mine. Though I did not know what to expect, I immensely enjoyed the conference. Not only was it well-organized, the topics were timely and vital to the community of higher education. I enjoyed each speaker and was particularly impressed with some of the dialogue in the chat sessions. Admittedly, I am not a multi-tasker, so the chat function in tandem with dialogue presented quite a challenge to me. And yet, I got a lot out of the opportunity, so I have to congratulate and thank The Chronicle of Higher Education for hosting such a worthwhile event.
My favorite section of the week was a dialogue between author Stephen Marche and Ian Wilhelm, an editor with The Chronicle of Higher Education. The conversation took so many positive and enlightening turns that I feel it necessary to focus solely on that twenty minute discussion in today’s blog.
During the past five to ten years, Stephen Marche has toyed with generative AI tools. As his skill with them increased, and as the AIs developed higher abilities, Marche began experimenting with longer works. He wrote one of his most recent books, Death of an Author, using prompts fed into three different AI tools. Marche developed a plot, something he says that AIs are not good at (yet), as well as some specifications about tone and style. The three different AIs then generated language which compose the bulk of Death of an Author. Obviously, this writing style opens the door to questions of authorship and originality. While he understands those questions, Marche still believes that Death of an Author is one hundred percent his. He explains that no one else would have made the choices that he made or fed in the same data, requirements, style, etc. So, while he did not write every word, he did direct the technology toward specific choices. Listening to his enthusiasm makes it hard not to be excited about this development. Clearly, he enjoyed the process.
He raised several key points for people who work in the humanities. First, Marche notes that “the distinction between understanding and production and the distinction between authorship and creative work … really matter.” While that may go without saying, it doesn’t change the fact that these terms are only vaguely understood. Authors often function behind the scenes only. We tend to naively assume that each book has one author, though that model has been faulty for a long time. For example, ghost writers have existed for decades, using the primary author’s name for leverage. James Patterson is a popular author who often employs ghost writers. Or, some writers collaborate. The author of the popular tween series Warriors is actually composed of three people who go by the pseudonym Erin Hunter. Furthermore, this question of authorship also makes translation a thorny issue. Do translators write, rewrite, repeat or own any of what they translate? Having never really resolved this issue, we arrive again at a point of misunderstanding authorship. But Marche’s infectious invitation to revisit that question asks us to come with an open mind and thirst for better understanding. I like this. I also tend to dislike strict adherence to categories, so the in-betweennes of something like translation or something like AI-generated writing intrigues me.
Next, Marche mentioned that the humanities have a crucial role to play as we move forward with AI generated texts. He says, “It actually matters that you understand the distinction between meaningful and truthful statements, between what authorial intent is and what authorial production is.” In my mind, we need to enhance critical thinking skills – something we often talk about but rarely see in student essays. Again, I like where this is headed, because clearly, discussion must be a more present feature of future education. In fact, discussion, or asking good questions, may be the key to learning. If we are expected to feed good prompts into generative AI models, then we might need to know more about the question and answer process itself.
He also made the point that writing like ChatGPT has already become an insult in popular culture, which means that writing in a vacuous, empty, vague (though correct) style is now worth much less than it was a few years ago. It’s boring writing, something a machine can do. Marche half-jokingly calls out the nature of academic writing by saying that it’s never been overly sought out. No one leaves college excited to write in an academic style. Maybe now we all must learn what it’s like to write like a human. If machines can do boring, technical, and empty, then humans must adjust with depth, insight, and creativity. Again, I like this line of thinking.
There’s no lamenting the fact of generative AIs because they are already here. I was particularly inspired by Stephen Marche’s insights, but also by his optimism. While change will likely be painful and slow, and I’m sure there will be bumps along the road, I want to agree with Marche that the future of writing can only benefit from AI. I love to think of the future in terms of positive change, so this will be the carrot I dangle at the end of the stick which will move myself and others towards a better understanding of AI generated texts. Furthermore, we have great opportunities to better understand how language actually functions. A rare opportunity for something that has existed for thousands of years.
While I am not as optimistic as Marche, I thoroughly enjoyed the discussion. For a full day’s events from ChronFest 2023, visit this link. Stephen Marche’s interview begins at two hours and twenty one minutes.