• Christopher McVey

    Christopher McVey is a senior lecturer in the College of Arts & Sciences Writing Program and Kilachand Honors College; he can be reached at cmcvey@bu.edu. Profile

Comments & Discussion

Boston University moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (EST) and can only accept comments written in English. Statistics or facts must include a citation or a link to the citation.

There are 10 comments on POV:人工智能正在改变大学写作。 Let’s Embrace It

  1. Hi, this is a very interesting article and I agree with the points made. To be honest, I have never heard of the use of artifical intelligence in writing until this article. However, I completely agree that writing classes should not just be about hte end product, but also the process in between. As I am taking WR415 here at BU this semester, I really appreciate the drafts, the peer edits, the conferences, and the evening events. We are doing different forms of writing such as Ted Talk and editing a wikipedia page, rather than traditional paper/research paper. It really makes me feel better about writing as a whole. Thank you for bringing this to light and I hope more professors and students can realize that writing classes should not be all about the end product.

    1. I also am in a WR415 class at BU this semester, and have never myself used an AI to help me write. To be honest, I don’t even know how to go about finding such a thing. I also really appreciate the diversity in the types of writing we do for the class, but honestly I am sometimes worried that I still am lacking the more “tradition” writing skills that AI seems to have mastered. Even though our WR415 class doesn’t emphasize or place importance on this style of traditional academic writing, so much of the rest of the world/academia still does. Hopefully that will change, but for now I still do have classes that require research papers and assume you know the language styles associated with stricter-form writing.

      This class has been my favorite writing class I’ve ever taken because of the emphasis it has placed on writing for the sake of writing and expression as supposed to writing that forces ideas to be crammed into the same paragraph structure and higher-ed vernacular. I hope that more writing within universities and professional settings embrace more open-form pieces. I can’t help but be pessimistic, however, and worry that I still would benefit from memorizing the rules of the 5-paragraph essay and “proper” (whatever that may mean) order of reasoning.

  2. First of all, it is surprising to know that artificial intelligence has already developed to such level that can assist students in college writing. Nevertheless, it is also not surprising since recently I learned that AI drawing existed and it is somehow similar to AI writing. In addition to writing plain words to form an essay to submit, to me a core part of writing consists of researching evidence and peer reviewing. AI may identify the theme or keywords and generate words in a specific genre, but sometimes people like to see words writing in a new style. The thing that separates AI and human is creativity. If we could encourage students to express creativity in writing courses, either through various media or different genre, then we shouldn’t be worrying about AI taking control of college writing courses.

  3. I think this is a fair point, but I disagree. A common mistake in analyzing AI advances is to assume that the limitations of current models are inherent, and not temporary. After the release of the first large language model, GPT-3, people noticed that eliciting high quality completions from the model was possible, but took many tries and substantial skill. This was dubbed “prompt engineering,” and commentators suggested that skill at writing itself would be supplanted with skill at prompt engineering. But this was disproven with the recent ChatGPT, where asking the model to perform tasks is straightforward and requires little careful wordplay. The original GPT-3, while capable, was never trained to be helpful or accurate, only to be good at predicting text completions – all it took to fix that was “incentivizing” it to (a technique called RLHF).

    Requiring students to, say, write about forms of media that current models cannot currently handle (video, podcasts, etc), is only a very short-term fix. This is in fact the next step in the roadmap for multiple major research groups, something that researchers in the field call “multimodality.” Rumors have it that GPT-3’s successor, GPT-4, will be multimodal, with a release window between December and February.

    Any other fixes will be similarly temporary. The explicit goal of OpenAI, DeepMind, and the collective billions of dollars pouring into the field, is the development of AI systems capable of carrying out all economically valuable human activity. I don’t expect them to halt progress when it’s no longer convenient for the rest of us.

    I hate to be so pessimistic, but I think the sad truth is that there is no easy fix. We are on the cusp of AI systems exceeding human performance across a wide swath of tasks, and it may simply be the case that writing as we know it today is not something humans do much of in the future – if we are even around at all.

    1. Hi Zach–

      I don’t expect them to halt progress, either. I fully expect that future iterations of this technology will become so fundamental to daily life that we’ll look back at this moment and laugh. We’ll say, “Remember how we thought it was so amazing that we could ask our Amazon Alexa device to turn on the lights or tell us the weather? Remember how we worried about social media disinformation campaigns run by /real humans/ sitting in a warehouse somewhere on the other side of the world, and not just by thousands of chatbots?”

      The point is not to develop “plagiarism proof” writing assignments; there has always been student cheating and there will always be student cheating, in one way or another. Rather, the point I hope to make is that we should seize this moment as an opportunity to think about what good pedagogy actually means: engaging with students not as workers who produce knowledge products for assessment and valuation, but rather as whole people worthy of full dialogue and intellectual exchange that can’t happen in “submit and done” writing assignments or traditional forms of grading. And I think this moment of disruption or change can be a real opportunity for positive transformation in the academy, as much as it presents new challenges.

  4. Hmm (sound of hard drive whirring). I wonder if AI-generated essays are inherently different from search-engine generated sources (MLA Bibliography; LC, ProQuest etc.). Or if search-engine generated searches are inherently different from print volumes of secondary criticism. Or if books themselves . . .

    The real and lasting threat to public education, it seems to me, is the very idea of a “classroom,” instead of an old Greek sitting down with two or three pupils and guiding them through a lifetime of instruction.

    In the short term, returning to the blue-book and pen will dispel most of the AI threat–if institutions are willing to provide adequate faculty to read what students actually can write when they sit down with pen and ink.

  5. He cites Microsoft as an investor in OpenAI: “So Microsoft is going to want to get its investment back by integrating these into Microsoft Word and into other tools. So you’re just going to see them used routinely, and it’s just going to be part of the writer’s repertoire”.

  6. I agree with many of the author’s points, especially with the content AI can produce that’s becoming more untraceable to any source. Rapid advancements in AI, particularly in natural language processing are leading to its increased integration into daily life. This integration is fueled by the growing computing power and more sophisticated models, resulting in automated tasks, improved decision-making, and even new creative ideas. AI-powered writing software is becoming more accessible, making it easier for people to generate content, assist with their writing, and even automate tasks. As AI continues to improve, it will become even more powerful and just as the author has mentioned, it will make some people to ”stop thinking” for themselves and ”simply rely on technological devices” to do their tasks.

Post a comment.

Your email address will not be published. Required fields are marked *