Aibots November 2026

aibots@lists.lysator.liu.se

2 participants
3 discussions

KernDev
by lihan62085＠emailkeep.com 03 Nov '26

03 Nov '26

KernDev is a software development company providing custom software, web and mobile applications, and AI-powered solutions. We help businesses address technical challenges through skilled development teams and practical technology services. Our goal is to build reliable software that supports business operations and long-term growth. For more information: https://kerndev.com/

1 0

KernDev
by lihan62085＠emailkeep.com 03 Nov '26

03 Nov '26

1 0

AI writing support in higher education: how do we assess student judgment?
by harri.evans90＠protonmail.com 01 Nov '26

01 Nov '26

Hello everyone, I would like to discuss how AI writing support can fit into university teaching while keeping student judgment central. My main question is how educators can distinguish an improvement in the submitted essay from an improvement in the student’s ability to reason, evaluate evidence, and revise independently. I am interested in examining EssaysBot’s approach through a practical educational question: can students explain and justify the revisions they make with AI assistance? For anyone who wants to examine the tool before joining the discussion, it is available at https://essaysbot.com/. I would focus the discussion on whether its output helps students recognize weaknesses in an argument, check supporting evidence, and make informed revision decisions. Starting with a specific writing problem Consider an illustrative undergraduate assignment about attendance requirements in university courses. A student has collected relevant readings but organized the essay around individual authors. The draft summarizes their positions without developing a clear argument. In this situation, I would begin with the assignment instructions and ask the student to explain the proposed thesis statement. Does it answer the research question? What evidence would support it? What reasonable objection would need to be addressed? These questions establish what the student understands before introducing a language model. They also help identify the appropriate task for a writing assistant. A difficulty with the central argument requires a different response from a difficulty with paragraph structure or reference formatting. For example, the student might propose that attendance requirements should differ between introductory seminars and independent research courses. I would treat this as a provisional claim to investigate. The research process would need to establish whether the available sources support the distinction and what qualifications it requires. Giving AI a bounded role Once the student has developed an initial position, I would consider using artificial intelligence to examine the outline. A focused prompt might ask which sections appear disconnected from the thesis or which assumptions require further evidence. I would ask for questions and explanations before requesting alternative prose. This keeps the activity close to guided practice: the student receives a response, examines its relevance, and decides what to do next. A generated draft can introduce additional problems if it becomes the organizing authority for the assignment. Its structure may appear coherent while overlooking the course objectives, misrepresenting a source, or making claims the student cannot substantiate. Output quality therefore needs to be assessed against the actual assignment and research notes. The practical boundary I would propose is simple: each use of a digital tool should address an identifiable writing problem. The student should be able to state that problem before using the tool and explain the resulting decision afterward. Making revision decisions visible Automated feedback becomes more useful when it addresses a specific concern. Asking whether a paragraph supports its topic sentence gives the writer something concrete to assess. A general instruction to “improve this essay” makes it harder to identify what changed and why. Suppose a paragraph promises a comparison but provides only description. The student could gather additional evidence, narrow the claim, or move the material to another section. Each option affects the argument differently. I would ask the student to select an option and explain the reasoning behind it. Accepting a recommendation should involve checking whether it preserves the intended meaning and the strength of the available evidence. A stylistic improvement can still remove an important qualification or imply an unsupported causal relationship. This feedback loop also creates a useful opportunity for assessment. Comparing two versions of a paragraph can reveal whether the student understands the difference between clearer expression and a substantive change in meaning. That explanation offers more information about learning than the revised paragraph alone. Keeping source evaluation within the workflow Citation and source quality require separate attention. A plausible reference does not establish that a publication exists, and correct formatting does not establish that the cited material supports the associated claim. I would expect the student to locate the original source, review the relevant passage, and check its context and limitations. Reference formatting should follow this substantive review. The same principle applies to plagiarism awareness. A paraphrase needs to preserve the source’s meaning, use sufficiently independent expression, and acknowledge the author. Changes in wording alone do not establish originality. Course requirements should also determine the permitted scope of assistance. Where disclosure is required, a brief record of prompts, outputs, and revision decisions can support an accurate account of the writing process. These expectations need to be clear before students begin drafting. Assessing what transfers to the next assignment As the deadline approaches, I would prioritize unsupported claims and inconsistent reasoning before editing and proofreading. Word count management should follow the same logic. A short essay may need fuller analysis; an overlong essay may need repeated material removed. My preferred closing activity would be to ask the student to explain one recommendation accepted and one rejected. I would then ask what they would check independently in their next essay. This connects the immediate revision cycle with the development of transferable writing skills. I would be interested in how others approach this in teaching, academic advising, or writing support. Do you assess students’ explanations of AI-assisted revisions? Have you found a manageable way to document those decisions without creating excessive administrative work? Which activities give you the clearest evidence that writing confidence is accompanied by stronger critical thinking?

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

Aibots November 2026