does the existence of a bunch of vapid mindless writing that has existed especially in the last 100 years (L Ron Hubbard's fiction in pulp magazines in the 30s comes to mind) just show AI will be more of the same? I feel like the real thing threatening the usefullness and existence of expertise and good writing and creative thinking isnt AI but economic conditions and social structures. in terms of not actually knowing local facts, I feel like the decline in local news coverage which has already happened has done more damage than AI ever will, but maybe im optimistic
And this is part of it, our knowledge repositories handed over to the vagaries of internet whims: "When the print materials are inaccessible because who needs human archivists anyway. When the print materials are destroyed because they take up too much space and are too costly to preserve. "
I guess the question is what the solution is here. The issue of schools not really testing students for competency is, as you note, not a new one, nor is it baked into generative AI. I have talked to plenty of Ivy League graduates whose reading comprehension is around a high school level (at best); I had one this weekend with a high school friend who graduated from Yale, yet chronically couldn't quote a statement accurately.
Yet schools continue to churn out graduates with very limited competency. I'm not sure how you teach otherwise or test competency better. The law school method of grading exclusively based on in class final exams with strict grading curves is perhaps one way, but how good of an assessment tool is that? And how do you do things like build writing skills over the course of an educational program without assigning writing (that can, in turn, be outsourced to GAI in the near future or, in the present and recent past, to willing parents or internet services)?
Seems like more of an existential educational issue than an AI-specific one.
That's the narrow problem. The big problem has nothing to do with students or their competencies. How do you teach writing in an infrastructure that might be at the edge of making writing of no relevance whatsoever, not by design but by accident?
I think we’re a long way off from that. Even if I didn’t think you were being a bit ungenerous to Shannara and Eragon, which are perfectly fine if somewhat cookie cutter works of fiction, I don’t think we’re anywhere close to AI being able to generate even minimally passable adult fiction. Or even to write decent nonfiction.
Generative AI is probably there when it comes to formatting documents, and is close to being able to do basic technical writing (for instance, a legal agreement). What I don’t think it may ever get to isn’t just sound analysis, but good style. Our most talented pop academic writers, like Jared Diamond and Paul Krugman, write with an engagement and clarity that I really don’t think you can teach a program to emulate, but which a well read and bright student can pick up, build on, and develop into their own.
I think we are already close to AI--and its funders--telling you that what AI wrote is passable, whether you like it or not, judge it so or not. ChatGPT is not an acceptable search engine, but we already have human beings tasked to produce journalism who are treating it as such. There's an existing layer of the cultural economy that can shovel whatever it has on hand into the maw and doesn't care much whether people who read or consume the product think much of it.
Depends by what standard. AI's funders can tell me what they want until they're blue in the face; if reality doesn't live up to their claims, it doesn't matter. What AI might be doing is replacing one form of popular mass drivel with another. An AI probably isn't far off from being able to do what Joe Rogan does on the radio. But there's also no real societal loss from replacing verbal diarrhea from a dim witted radio host with verbal diarrhea from a large language model.
does the existence of a bunch of vapid mindless writing that has existed especially in the last 100 years (L Ron Hubbard's fiction in pulp magazines in the 30s comes to mind) just show AI will be more of the same? I feel like the real thing threatening the usefullness and existence of expertise and good writing and creative thinking isnt AI but economic conditions and social structures. in terms of not actually knowing local facts, I feel like the decline in local news coverage which has already happened has done more damage than AI ever will, but maybe im optimistic
And this is part of it, our knowledge repositories handed over to the vagaries of internet whims: "When the print materials are inaccessible because who needs human archivists anyway. When the print materials are destroyed because they take up too much space and are too costly to preserve. "
I guess the question is what the solution is here. The issue of schools not really testing students for competency is, as you note, not a new one, nor is it baked into generative AI. I have talked to plenty of Ivy League graduates whose reading comprehension is around a high school level (at best); I had one this weekend with a high school friend who graduated from Yale, yet chronically couldn't quote a statement accurately.
Yet schools continue to churn out graduates with very limited competency. I'm not sure how you teach otherwise or test competency better. The law school method of grading exclusively based on in class final exams with strict grading curves is perhaps one way, but how good of an assessment tool is that? And how do you do things like build writing skills over the course of an educational program without assigning writing (that can, in turn, be outsourced to GAI in the near future or, in the present and recent past, to willing parents or internet services)?
Seems like more of an existential educational issue than an AI-specific one.
That's the narrow problem. The big problem has nothing to do with students or their competencies. How do you teach writing in an infrastructure that might be at the edge of making writing of no relevance whatsoever, not by design but by accident?
I think we’re a long way off from that. Even if I didn’t think you were being a bit ungenerous to Shannara and Eragon, which are perfectly fine if somewhat cookie cutter works of fiction, I don’t think we’re anywhere close to AI being able to generate even minimally passable adult fiction. Or even to write decent nonfiction.
Generative AI is probably there when it comes to formatting documents, and is close to being able to do basic technical writing (for instance, a legal agreement). What I don’t think it may ever get to isn’t just sound analysis, but good style. Our most talented pop academic writers, like Jared Diamond and Paul Krugman, write with an engagement and clarity that I really don’t think you can teach a program to emulate, but which a well read and bright student can pick up, build on, and develop into their own.
I think we are already close to AI--and its funders--telling you that what AI wrote is passable, whether you like it or not, judge it so or not. ChatGPT is not an acceptable search engine, but we already have human beings tasked to produce journalism who are treating it as such. There's an existing layer of the cultural economy that can shovel whatever it has on hand into the maw and doesn't care much whether people who read or consume the product think much of it.
Depends by what standard. AI's funders can tell me what they want until they're blue in the face; if reality doesn't live up to their claims, it doesn't matter. What AI might be doing is replacing one form of popular mass drivel with another. An AI probably isn't far off from being able to do what Joe Rogan does on the radio. But there's also no real societal loss from replacing verbal diarrhea from a dim witted radio host with verbal diarrhea from a large language model.