Overeducated, Ph.D.

Writings, Regrets, and Re-skillings during the AI Revolution

Author: Professor Lamb

  • Neurodivergent Friendly Teaching 4: Participation Is Not a Speed Contest

    I used to grade participation the way I was graded: who spoke, how often, how quickly. It took me longer than I would like to admit to notice what that actually measured. It measured comfort with speaking, processing speed, and confidence, none of which was on the syllabus.

    The student who answers in the first two seconds is often not the student who understands best. Sometimes they are simply the student least troubled by being wrong. Meanwhile, the student who needs ten seconds to organize a thought, the one who is translating from another language, and the one who finds eye contact physically uncomfortable are all being quietly marked down for how their brains and bodies work.

    Here is what I changed.

    I build in thinking time before discussion. The classic research on wait time found that most instructors leave under a second between a question and calling on someone. Stretching that to a deliberate pause changes who answers and how well. I go further and ask everyone to write for two minutes first. Writing first means the fast talkers have something more considered to say and the slow processors have something to say at all.

    Where a written response meets the objective, I accept it as full participation. If the goal is to show that you understand why a routing loop forms, a paragraph in the LMS discussion shows that as well as saying it aloud, and often better. If the goal is oral explanation because the course teaches client communication, then I keep the spoken requirement and scaffold it, with the question posted in advance and a chance to rehearse in pairs.

    Group work gets explicit roles and a named deliverable. “Discuss in groups” is an invitation for the loudest person to run the show and for others to disappear. “One of you drives, one navigates, one records the steps, and you submit the recorded steps at the end” gives every student a defined job. Pair programming already works this way for good reason. Roles rotate so nobody is permanently the scribe.

    And I have stopped reading faces as evidence. A student looking at the desk may be listening harder than the one nodding at me. Eye contact is culturally variable and, for many autistic people, costs attention that would otherwise go to the content. Quick speech is a trait, not a competency. What I want is the thinking, and I have learned to build several doors into it.

    For a community college, this matters more than it might elsewhere. Our rooms hold nineteen-year-olds and fifty-year-olds, people returning after a long time away, and people who have never once been asked their opinion in a classroom. Offering multiple workable ways to take part is not lowering the bar. It is making sure the bar measures the thing I said it was measuring.

  • Neurodivergent Friendly Teaching 3: One Thing at a Time in the Lab

    Picture a typical software demonstration. I am at the front, clicking through a configuration. Students are watching the projector, listening to me narrate, reading the text on the screen, and trying to reproduce each step on their own machine. That is four channels of input and one channel of output, all running at the same time, and every one of them is competing for the same limited attention.

    Cognitive load research calls this the split-attention effect, and it hurts everyone. But it does not hurt everyone equally. A student with ADHD who loses the thread for eight seconds is now three steps behind with no way back in. A student who processes spoken language a beat slower is always translating what I said a moment ago while I am already saying the next thing. A student with a visual processing difference is squinting at a cluttered slide. By minute ten, a third of the room is quietly lost and too embarrassed to say so.

    The fix is to reduce the number of demands competing at any one moment.

    First, put the instructions out before class, in an accessible format. A student who has already skimmed the step list walks in knowing the shape of the hour. Some of my students read it on the bus. Some run it through a screen reader. Either way they arrive with a scaffold rather than a blank.

    Second, alternate watching and doing. I demonstrate three or four steps while students only watch, hands off the keyboard. Then I stop, and they do those same steps while I walk the room. Then we repeat. Nobody has to watch and type simultaneously, and the walk gives me a real read on who is stuck.

    Third, keep a short step list visible the whole time, on a second screen, a whiteboard, or a printed half-sheet. When a student drops the thread, the list is the way back in. It should be terse: numbered, one action per line, no paragraphs.

    Fourth, caption everything you record and clean up everything you project. Captions are not only for deaf and hard-of-hearing students; they help the student in a loud house, the second-language learner, and anyone whose auditory attention wanders. Slides with one idea and generous white space get read. Slides with nine bullets get ignored.

    Community college labs make this harder than a lecture hall would. Machines vary, students arrive at different points, and there is rarely a teaching assistant to catch the person who fell behind. That is precisely why the structure matters. The alternating rhythm and the visible list are doing the work an assistant would do. Fewer competing demands means more students finishing the lab, and it means I spend the period teaching rather than re-teaching the same three steps to one row at a time.

  • Neurodivergent Friendly Teaching 2: Be Explicit

    The most common question in my inbox is not about programming code or subnetting. It is some version of “what are we supposed to do?” My personal favorites: “What did I miss? Why is my grade zero? Why am I failing your class?” For years I read that as students not reading. I now read it as an opportunity for connection through using my words.

    Academic structures and course assignments that are clear to me, war-torn veteran of decades of higher-ed, are sometimes unclear to a student, especially for those who are taking their first college course online. For many students with ADHD, dyslexia, autism, et al, a vague task is not only irritating. It can be the wall of inertia that ends semesters before they start. Executive function (the mental machinery for getting started, sequencing, and finishing) is exactly what a fuzzy prompt taxes the most, which can trigger negative affective ripples across the whole person.

    So now every Activity I require explicitly states five things, in the same “P-F-O-D-E” order, every time:

    1 The Purpose (Why): why this exists and what skill it builds. If I can’t say it in one sentence, it’s a sign I have not performed enough executive function myself as the instructor and that I need to have some alone prep time.

    2 The First Step (How): the very first thing to do, physically. “Open Packet Tracer and load the .pka file you downloaded from Module 6.” Starting is the hardest step for many people, and naming it removes the blank page.

    3 The Outcome (What): or expected deliverable, what submission gets turned in, in what format, with what file name. A screenshot? A .java file? A paragraph in the text box?

    4 The Due Date (When): date, time, and whether late work is accepted.

    5 The Evaluation (Who): what successful students do who earn the highest grades, a plain-language description or a short exemplar. Not a rubric with twelve rows, though a rubric can sit underneath. Two or three sentences a student can check their own work against.

    Note that my instructions live in one predictable place in the LMS, and only there. For me that is the assignment page itself, linked from the week’s module, laid out identically every week. Not the announcement, not the slide deck, not a reply buried in a discussion thread. When a student knows where to look, the looking stops costing them anything.

    None of this is an accommodation for some students. It is universal design for all my students: a change to the default that removes a barrier for some students and costs nothing for the rest. The strong student skims it in ten seconds. The student who would have emailed me at 11 p.m. does not need to. Both of them get to spend their effort on the actual work.

  • Neurodivergent Friendly Teaching: Introduction

    This post starts a short series on neurodivergence in higher ed teaching and community colleges. Half of the posts are about the students in front of us. The other half is about those of us at the front of the room whose own brains do not run on the syllabus schedule nor follow the lesson plan as much as we might want them to.

    Neurodiversity is the broader umbrella term. It describes the plain fact that human brains vary in how they attend to, remember, process sensory input, read, and communicate, and it treats that variation as natural rather than as a list of defects. The term was deliberately modeled on biodiversity: the population is neurodiverse. A single person is not, any more than one tree is a forest.

    ‘Neurodivergent’ is the word for an individual whose cognition diverges from what a given society treats as ‘normal.’ It is an informal family rather than formal diagnosis, commonly including autism, ADHD, dyslexia, dyscalculia, dyspraxia, and Tourette’s; I include OCD, Traumatic brain injury, and other mental health conditions as well. The counterpart is ‘neurotypical.’ Neither term comes from a clinic. They come from advocacy communities, and their edges are interpretive.

    Neurodivergence does not imply one learning profile. Two autistic students in the same section may need opposite things. One wants the discussion question in writing an hour early; the other wants to be left alone until the lab starts. So the useful question is not “what do all neurodivergent students need?” but rather “what helps this particular student succeed?” asked directly of and answered by that student at some point.

    Let me be explicit about what I mean. I am not saying that every student is neurodivergent these days. Being an adult learner, learning in a second language, feeling stressed, or arriving exhausted from an eight-hour shift does not itself make someone neurodivergent. Those circumstances can create barriers to learning, and some of the same teaching choices can help. Overlapping support needs do not make them the same condition.

    We do have evidence that more students are being identified with neurodivergent conditions. A large study of college students from 2009 to 2015 found increasing reports of ADHD diagnosis or treatment. That is evidence of growth in identified needs, although it does not tell us how much comes from changes in recognition, disclosure, or underlying prevalence. I suspect the proportion of neurodivergent students may be especially high in community colleges, but I regard that as a working hypothesis rather than an established comparison. Either way, these students are a regular part of our classrooms, and planning for their presence should be ordinary teaching practice.

    The practical issue for me is scale. Community college teaching comes with high demand and a heavy teaching load. I have hundreds of active students at any one time. I cannot build a separate version of every lesson around every student’s diagnosis, and I cannot make useful support depend on knowing who has one. I can build clear instructions, predictable organization, accessible materials, manageable sequences, and appropriate choices into the course from the beginning, then respond to individual needs as they arise.

    In that sense, I teach as if the whole classroom were neurodivergent: I anticipate barriers involving attention, working memory, sensory processing, reading, and communication throughout the design. That is a planning stance, not a diagnosis of the room. Universal Design for Learning (UDL) gives me a way to make those supports and choices available to everyone. A clear first step can help a student with ADHD get started and help a tired student find their place. Captions can help students with different hearing, language, or attention needs. Every student can benefit from that design, without every student needing the same support or benefiting from the same option. I still ask what helps, adjust where I can, and honor individual accommodations.

    The answer is not heroics. It is simple design choices that recognize learner variability while keeping teaching sustainable for the person doing it, because an exhausted instructor helps nobody. Your own experiences with a neurodivergent brain, whatever it is, can boost empathy but do not necessarily predict (or even resemble) your student’s process.

    First up are five posts on helping students with neurodivergent minds with concrete pedagogical recommendations regarding explicit tasks, competing demands, participation, sensory needs, and authentic assessment. Then come five more on teaching in the classroom with neurodivergent mind and what (sometimes) helps: cue sheets, live multitasking, energy, workplace access, and partial disclosure.

  • ChatGPT at Work 11: How Not to Get Fired Using AI

    Experimenting with AI can be great fun. Generate a ridiculous image, rewrite a paragraph, build a little application. When a harmless personal experiment fails, you can often shrug and try again. At work, the consequences may belong to other people. The spreadsheet contains customer information. The document describes an internal investigation. The chatbot’s recommendation affects an employee’s access to a system. You may be able to open those files without having permission to upload them somewhere else.

    That is the first workplace rule: access to information is not permission to share it with an AI service. Before uploading anything, find out which tools, accounts, and uses your organization has approved. Your personal paid subscription does not become an approved business system because you bought the expensive model. Enthusiasm is not an authorization level.

    Account settings matter, too. In personal ChatGPT accounts, you can turn off “Improve the model for everyone” under Data Controls to prevent your conversations from being used for training. OpenAI excludes business data from training by default in ChatGPT Business, Enterprise, and Edu. Those are meaningful distinctions, but opting out of training does not establish that your employer permits the upload. Personal data controls and business data protections answer different parts of that question.

    Training, storage, access, and monitoring are separate issues. A conversation can be excluded from model training and still be retained under a workplace policy. Eligible Enterprise and Edu workspaces support access to certain conversation records through compliance tools; exactly what is available depends on the product and configuration. Treat a workplace chat as a work record, rather than assuming the conversational interface makes it private. OpenAI’s administrative guidance explains those distinctions.

    Then there is the output. A model can produce a polished explanation with an invented source, a wrong calculation, or an assumption that makes no sense in your organization. Your name may go on the report. The chatbot will not be attending the meeting where you explain it. For an IT example, imagine asking ChatGPT to help prioritize support tickets. Use an approved environment and provide only the information needed for the task. Check the proposed priorities against the actual evidence. Let it draft a recommendation; require the appropriate person to approve actions that disable accounts, reset credentials, or isolate devices.

    Be equally explicit about connected tools: “Draft the email, but do not send it.” “Review the records, but do not change them.” Use available permission controls to enforce those boundaries as well as stating them in the prompt.

    The useful workplace skill is knowing what you may delegate, what information you may provide, and what you must check yourself. If the policy is unclear, ask the person responsible before experimenting with real organizational data. “I was trying to save time” is a much better opening to a success story than an incident report.

  • ChatGPT at Work 10: Your Files Are Part of the Prompt

    In a demonstration I recorded for class, I gave ChatGPT a short assignment: create a seven-slide networking presentation for a fictional company called Country Garden Supply. Include a title slide, a network diagram, and static IPv4 addresses.

    That sounds like an underspecified prompt. Where were the business requirements? What was the assignment? How would the model know what belonged in the presentation?

    The answers were in the project folder. Before assigning the presentation, I asked ChatGPT what the final project required. It found and read an assignment page I had saved as an HTML file in my Intro to Networking folder. That document supplied details I did not have to dictate again.

    Local projects in the ChatGPT desktop app can give the system access to folders on your computer. Those files become available for searching, reading, and editing. Access does not mean every document has already been read or that the correct version will always be selected. OpenAI’s project documentation explains how this works.

    Watching the demonstration later, I noticed myself saying that I had barely provided any context. But I had. I had prepared the folder.

    That is a useful shift in how we think about prompting. The request in the chat box is only part of the assignment. The requirements document, spreadsheet, template, and reference materials can carry much of the detail.

    For business work, prepare those materials deliberately. Identify the current policy. Label the approved template. Separate old drafts from the version that should govern the task. Tell the agent which source takes precedence when documents disagree.

    Then ask it to summarize the requirements before it starts building. That gives you an early opportunity to catch the wrong file, an outdated instruction, or a missing constraint. The next improvement to your AI workflow is fifteen minutes spent organizing the co-work folder.

  • ChatGPT at Work 9: GPT-Live Can Listen While It Talks

    What better way to learn about chatbots than to interview one? For my AI Tools for Business class, I asked ChatGPT to play the subject matter expert. We discussed customer service, helpdesks, business costs, and the risks of letting a confident machine represent your organization.

    But the interface was part of the lesson. I was using GPT-Live, OpenAI’s new voice model family, introduced in July 2026. Its defining feature is full-duplex conversation: it can listen and speak at the same time. The two versions introduced were GPT-Live-1 and GPT-Live-1 mini. OpenAI’s announcement explains the change.

    That sounds like a small improvement until you think about how people actually talk. We pause without surrendering the floor. We offer little acknowledgments while somebody else speaks. We interrupt because we missed a word, disagree with an assumption, or need the explanation to slow down. Sometimes we need a moment to find the phrase we wanted. My students are familiar with this particular feature of their instructor.

    Earlier voice systems depended on detecting when you had finished speaking. Guess too early and the machine cuts you off. Guess too late and everybody waits. GPT-Live continuously processes incoming audio while generating speech, allowing the model to decide whether to speak, listen, or pause. OpenAI’s engineering account describes how it removed the separate turn detector from the live audio path.

    In my demonstration, I stumbled over “by the dozen” and “in droves.” The chatbot supplied the phrase with the hilarious aside “English. Lovely language.” LOL. Later, the audio broke up during an explanation of chatbot return on investment. I interrupted and asked it to start again at a more beginner-friendly level. It did. The technology still had an imperfect moment, but conversation gave me a straightforward way to recover.

    There is another interesting piece behind the voice. GPT-Live can delegate harder reasoning and searches to another model while remaining available for conversation. The system separates the immediate demands of talking from work that takes longer. That helps explain how a responsive voice interface can connect to more substantial AI capabilities. OpenAI’s architecture explanation covers this division of labor.

    For work and learning, I see possibilities in rehearsing a presentation, practicing a difficult customer conversation, or talking through an unfamiliar concept. You can steer as you go: “Give me an example.” “That assumes too much background.” “Wait, let me finish.”

    The charm deserves scrutiny, too. GPT-Live makes talking to a computer feel more natural. We still have to decide whether the computer has said anything worth believing. A fluent explanation can make it sound authoritative. Neither establishes that its answer is correct. In the interview, I assigned the chatbot the role of expert; that assignment did not give it credentials. My practical suggestion is to try a short conversation about something you know well. Interrupt, change direction, ask for clarification, and check its claims. Judge both the interaction and the information.

  • ChatGPT at Work 8: To Chat or To Work? That Is The Question

    ChatGPT now asks us to make a decision before we ask it to help us make decisions: to Chat or to Work? That is the question! Apparently even the chatbot needs to know whether this conversation could have been a one-line email.

    The basic rule of thumb: Use Chat to think with ChatGPT. Use Work to delegate a job to ChatGPT. Chat is useful when you want an explanation, a brainstorm, a comparison, or help finding the right words. You are directing the conversation as it develops. Work is designed for a larger assignment with an outcome you can deliver and share: a report, a presentation, a spreadsheet, or even an application prototype/demo. That is the practical distinction in OpenAI’s guidance.

    “Help me think through an employee onboarding session” belongs comfortably in Chat. You can discuss the audience, reject boring activities, and figure out what people actually need to learn. “Use these policies and notes to create the onboarding presentation, check that every required topic is covered, and return an editable PowerPoint” is a Work assignment. The useful distinction is how much of the process you want to hand over. Both modes use the same model. So what actually changes?

    The (multimodal) Large Language Model (LLM) is only part of the system. The software program around it determines which tools it can use, what information it can retain during a task, and how it continues after an intermediate result. Developers call this surrounding machinery an agent harness. An agent loop can repeatedly ask the model what to do next, execute a tool call, return the result, and continue until it reaches a stopping point. Work provides an environment for carrying out an assignment across multiple steps. Cloud Work can continue supported tasks while you are away; local Work can use files and applications made available through the desktop app.

    When the assignment is fuzzy, talk it through with Chat. Once you can describe the deliverable and what would make it acceptable, Work becomes easier to direct and evaluate. OpenAI recommends reserving Work for substantial tasks and using Chat for quick questions and short rewrites. Delegation still leaves you with a job. Review the data sources, check and improve output quality, and decide whether the result meets your requirements. Choose Chat when you want to work through the thinking together. Choose Work when you can assign the job, have clear requirements, and intend to eventually share an artifact with others.

  • ChatGPT at Work 7: Editable Writing Blocks

    Editable writing blocks may be the best thing since sliced bread. That sounds like a ridiculous amount of enthusiasm for a text Edit box, but the feature removes one of the most persistent frustrations of writing with ChatGPT: the awkward distance between discussing a document and actually editing it.

    OpenAI’s older Canvas interface pointed in the right direction. A longer draft opened in a separate pane, giving the document more room and providing editing tools outside the ordinary chat stream. It was useful, but it also felt clunky. The conversation lived in one place and the document lived in another. Revising often felt like stepping out of the discussion and into a second workspace.

    The newer writing block, with its Edit control, is far more natural. The document appears directly inside the conversation. I can ask ChatGPT for a draft, click Edit, rewrite a sentence myself, delete an awkward paragraph, or add a detail that only I know. Then I can continue the conversation from the version I actually edited. There is no need to copy the entire document into another program or paste the revised version into a new prompt. This makes iteration feel less like repeatedly ordering new drafts and more like genuine collaborative editing. I can preserve what works, fix small problems myself, and ask ChatGPT to focus on the parts that still need help.

    For an IT example, put a draft network-change plan in an editable writing block. Correct the device names, IP addresses, maintenance window, and rollback steps directly. Then ask ChatGPT to review the edited plan for unsupported assumptions, missing validation steps, unclear instructions, and anything that would confuse a junior technician. The human supplies the operational truth; ChatGPT helps inspect and improve the document.

    The first response is still only a draft. The difference is that the draft no longer feels trapped inside a chat response or exiled to a separate Canvas. The writing block keeps the document and the conversation together. That small interface change makes revision faster, clearer, and much more pleasant—and keeps human judgment in control.

  • ChatGPT at Work 6: The First Response Is Slop

    The first response is slop. It is not the one to publish or share. It is a shot in the dark. It must be improved, edited, and/or regenerated with better prompting, and often by adding more context, before letting anybody else see it.

    Iteration is the difference between AI slop and AI-human quality. The latter is the one that genuinely saves time and advances work tasks. After the first response, pause, tell ChatGPT exactly what you like and dislike. “The opening works, but the middle sounds inflated.” “Keep the example and replace the conclusion.” “This is too polite. Say what the problem actually is.” The model needs your expert steering and context awareness. If you find yourself correcting ChatGPT the same way multiple times, consider adding it to your Custom Instructions (in Settings -> Personalization). Your future self will thank you.

    Ask for multiple tries when you do not yet know what quality looks like. Generative AI is based on statistical probabilities so even the exact same request will produce different results, some better than others. Or you can request three genuinely different versions, structures, or approaches—not three lightly rearranged copies of the same draft. Compare them. Keep one opening, another explanation, and a third conclusion. A failed attempt can still show you what you do not want.

    When a response misses, do not merely say, “Try again.” Name the failure. Identify the requirement it ignored, the assumption it invented, the tone it misunderstood, or the detail it must preserve. If part of the answer is correct, say so. Otherwise, the next attempt may repair one section by breaking another.

    Sometimes the conversation gets stuck in a groove. Stop. Wait. Read the result later with fresh eyes. Start a new chat with a cleaner prompt, provide a better example, or change the order of the instructions. You can also try a different model. A more capable model may help with difficult reasoning or synthesis; another model may simply approach the wording or structure differently. Compare the results against your actual requirements rather than assuming that newer, larger, or slower thinking automatically means better.

    Iteration is not endless polishing. Watch out for AI’s tendency to worsen your perfectionism. Short and authentic beats ‘shock and awe’ almost every time. Imperfection can make you sound more human and real. So keep the loop of iteration/improvement/regeneration short, tidy and still human. Experiment, change one or two things, and try again. Done. Eventually the document meets the requirements, the remaining flaws do not matter, or more revisions stop contributing value.