Try running the draft through Claude. Ask for an assessment of its coherence. I would be interested in hearing the results of this experiment. Claude won’t do anything to it or with it. Also I’m curious about how you determine an “AI expert.”
Tom, just try it. I think you’ll be surprised. I just don’t understand the fear of trying. I’m not sure you have the full picture of computational language and the role of syntactic trees. I can tell you from personal experience Claude is an excellent tool for assessing coherence. I wouldn’t value ChatGPT for the coherence check. Its training corpus leans more heavily a traditional essayist tradition. Yes, most LLMs are trained on Western texts in English and Lord knows the bias that infests the writing of mortals in the Western tradition. Racism is real in the world and AI is mirroring this world. That doesn’t mean Claude can’t do a coherence assessment. Ignore it if you don’t agree. But I would be surprised if YOU aren’t pleasantly surprised at the extra set of contacts Claude (I would say “eyes” but that’s anthropomorphizing the machine. Btw, Claude isn’t racist any more than a shovel or a Tesla. Claude is not human. The racism in the synthetic output shows up because of all the human-written racist texts on the Internet. My advice is to keep an open mind. Don’t get attached to AI ideologues. I have a doctorate in language and literacy, and most of what I read from these experts is speculative and based an anecdotes. Try your post with Claude. Copy the entire text, paste it into Claude’s window, write “Check this piece for coherence. I needs a clear through line, but there are a couple of places after the pace picks up that might be speed bumps for my reader. I don’t need perfection, but everything needs to flow smoothly.” I’ll bet you a Snickers bar you’ll be surprised. Somebody near you has a $20 per month Claude Pro Sonnet not. You may not be so impressed with other bots. All the best to you! High school teachers are heroes in my book.
Thank you for your comment. I apply what I call the AI Resource Test to using generative AI. For LLMs it is, "Are the prompts and generated text worth a bottle of water?" We know that LLMs consume water in under-resourced communities. It makes me very reserved in my use of this tool. I wrote about it here: https://www.criticalinkling.com/i/146245711/apply-the-ai-resource-test
Try running the draft through Claude. Ask for an assessment of its coherence. I would be interested in hearing the results of this experiment. Claude won’t do anything to it or with it. Also I’m curious about how you determine an “AI expert.”
While generative AI is good at word form and grammar, I am very hesitant to use it for any deeper feedback than that. Here is one such example of why: https://melissa-warr.com/genai-is-racist-period/. "AI expert" is subjective. I judge by their work. Here is a link to some experts I trust: https://www.criticalinkling.com/p/teachers-follow-these-experts-to-learn-ai
Tom, just try it. I think you’ll be surprised. I just don’t understand the fear of trying. I’m not sure you have the full picture of computational language and the role of syntactic trees. I can tell you from personal experience Claude is an excellent tool for assessing coherence. I wouldn’t value ChatGPT for the coherence check. Its training corpus leans more heavily a traditional essayist tradition. Yes, most LLMs are trained on Western texts in English and Lord knows the bias that infests the writing of mortals in the Western tradition. Racism is real in the world and AI is mirroring this world. That doesn’t mean Claude can’t do a coherence assessment. Ignore it if you don’t agree. But I would be surprised if YOU aren’t pleasantly surprised at the extra set of contacts Claude (I would say “eyes” but that’s anthropomorphizing the machine. Btw, Claude isn’t racist any more than a shovel or a Tesla. Claude is not human. The racism in the synthetic output shows up because of all the human-written racist texts on the Internet. My advice is to keep an open mind. Don’t get attached to AI ideologues. I have a doctorate in language and literacy, and most of what I read from these experts is speculative and based an anecdotes. Try your post with Claude. Copy the entire text, paste it into Claude’s window, write “Check this piece for coherence. I needs a clear through line, but there are a couple of places after the pace picks up that might be speed bumps for my reader. I don’t need perfection, but everything needs to flow smoothly.” I’ll bet you a Snickers bar you’ll be surprised. Somebody near you has a $20 per month Claude Pro Sonnet not. You may not be so impressed with other bots. All the best to you! High school teachers are heroes in my book.
Thank you for your comment. I apply what I call the AI Resource Test to using generative AI. For LLMs it is, "Are the prompts and generated text worth a bottle of water?" We know that LLMs consume water in under-resourced communities. It makes me very reserved in my use of this tool. I wrote about it here: https://www.criticalinkling.com/i/146245711/apply-the-ai-resource-test