Written communication for a specific audience and a specific purpose entails careful attention to content that fulfills the purpose, to inferences that can safely go unstated, to prior knowledge which may need building or refreshing, to emphasis on important points, to tone and register and persona, to emotional resonance, word choice, and a range of genre-related, often ideological concerns.
Thanks Terry for your thoughtful essay here. When you said, "LLMs simulate human language the way cubic zirconia simulates diamonds", making a case that these are language machines and not intelligence machines, I found the comparison helpful. I may use this with the teachers I am coaching right now around assessment, A.I., and writing.
Terry - high school ELA teacher here - your suspicion is accurate about the "Language Arts" in at least the three public schools I have taught at in Western PA. Not a fixed truth, but certainly that is the sort of systematic entropy that can set in for the instructor and students. Great entry here from you! I emphasize embodied, disembodied, and re-embodied ethics at the core of the AI and Ethics class I taught last semester, and also note early and repeat somewhat often what you emphasized about the use of the phrase "artificial intelligence" - though I start at the Dartmouth conference a year earlier so I can emphasize that these "language models" are really more math computations from mathematicians. How much direct mathematical computation do I want in my students' writing? How much math, computational or not, do most readers want in their poetry or prose or plays or short stories? Sure., there is syntax...but maybe E.E. Cummings had a point about feelings.
Thanks, Scott. I appreciate the validation of my intuition about ELA instruction and your experiences in teaching AI and ethics. The history point is excellent. Tracing back to the 1955 Dartmouth proposal helps students see that what we call "AI" began primarily as mathematical and computational exercises, not as the almost mystical force it's sometimes portrayed as today. Computational language is an oxymoron in my mind as well. There's more authentic language in a chimpanzee than in an LLM eh? Cummings knew that breaking rules could sometimes communicate feelings more authentically than following them. Maybe that's a valuable lesson for thinking about both writing and technology? Great insight into a framework for AI and Ethics!
Thanks Terry for your thoughtful essay here. When you said, "LLMs simulate human language the way cubic zirconia simulates diamonds", making a case that these are language machines and not intelligence machines, I found the comparison helpful. I may use this with the teachers I am coaching right now around assessment, A.I., and writing.
Terry - high school ELA teacher here - your suspicion is accurate about the "Language Arts" in at least the three public schools I have taught at in Western PA. Not a fixed truth, but certainly that is the sort of systematic entropy that can set in for the instructor and students. Great entry here from you! I emphasize embodied, disembodied, and re-embodied ethics at the core of the AI and Ethics class I taught last semester, and also note early and repeat somewhat often what you emphasized about the use of the phrase "artificial intelligence" - though I start at the Dartmouth conference a year earlier so I can emphasize that these "language models" are really more math computations from mathematicians. How much direct mathematical computation do I want in my students' writing? How much math, computational or not, do most readers want in their poetry or prose or plays or short stories? Sure., there is syntax...but maybe E.E. Cummings had a point about feelings.
Thanks, Scott. I appreciate the validation of my intuition about ELA instruction and your experiences in teaching AI and ethics. The history point is excellent. Tracing back to the 1955 Dartmouth proposal helps students see that what we call "AI" began primarily as mathematical and computational exercises, not as the almost mystical force it's sometimes portrayed as today. Computational language is an oxymoron in my mind as well. There's more authentic language in a chimpanzee than in an LLM eh? Cummings knew that breaking rules could sometimes communicate feelings more authentically than following them. Maybe that's a valuable lesson for thinking about both writing and technology? Great insight into a framework for AI and Ethics!