AI-Powered Assessments in Education

𝗧𝗵𝗶𝘀 𝗜𝘀 𝗛𝗼𝘄 𝗬𝗼𝘂 𝗧𝗿𝗮𝗶𝗻 𝗬𝗼𝘂𝗿 𝗔𝗜 You want to use AI to make lesson planning easier. But generic AI outputs can waste more time than they save. To get the most out of AI, you need to teach it your way of teaching first. You do this by writing down your core teaching principles. What do you believe in? What materials do you use? What is your practice philosophy? - Write down 3-5 short principles that guide your teaching - Identify your most-used materials and method books - Tag key pages in your method books with specific skills and concepts This helps the AI understand your approach and generate plans that fit your style. Start with your top 50 pieces and 2-3 core method books. Use a template to organize your pieces by composer or style. When you configure your AI tool with this information, it can generate plans that you would write yourself. This saves you hours of time and preserves your unique teaching voice. Source: https://lnkd.in/grtrJncY Optional learning community: https://t.me/GyaanSetuAi

For the longest time, I kept hearing this simple narrative: students are using AI to cheat, to shortcut thinking, to let machines do the work.☝🏻 But the more I read this recent study, the more that idea feels incomplete. What I found interesting is that students aren’t just handing over their writing to AI. They’re negotiating with it. They’re using it to get unstuck, to brainstorm, to shape rough ideas, and then stepping back in to actually write.🤯 That feels very human. When I think about my own process, it’s rarely linear. I don’t sit down with perfect clarity and just produce something meaningful. I struggle, I explore, I rewrite, I question my own thoughts. And in that sense, AI seems to be slipping into the process not as a replacement, but as a kind of thinking partner. What stood out to me most is that students often don’t accept AI output as it is. They edit it, challenge it, sometimes reject it completely when it feels too generic or off-track. That tells me something important: authorship hasn’t disappeared, it’s just evolving.👏🏻 But there’s also a subtle risk here. If AI becomes the default starting point, do we slowly lose the discomfort of beginning from scratch? Because that blank page, as frustrating as it is, is also where original thinking usually starts.🤔 I don’t think this is a story about dependency yet. It feels more like a transition phase. Where we’re all, students included, trying to figure out how much of our thinking we want to outsource, and how much we want to hold onto. And maybe that balance will define the kind of thinkers we become.🤷🏻♂️ #AI #Education #Writing #FutureOfWork

This hits. As an Instructional Designer who has embedded AI deeply into my workflow, I see this every day: "Generating is easier now. Thinking is not." AI can help produce faster, but the real value is still in the human work: - asking better questions - spotting weak logic - shaping the experience, and - refining what actually matters That is why Bloom’s Taxonomy feels worth revisiting for the AI age. Creating is no longer the hard part. Judgment is. Well worth the read. https://lnkd.in/eJeYPYtQ

The decision isn’t “Should I use AI?” It’s “Where?” I really appreciate this framing. It feels especially useful in higher education, where the question is not simply whether AI belongs in our work, but where it adds value, where it creates risk, and where it may interfere with the learning we are actually trying to support. That is something I will be bringing into the AI workshops I offer to faculty. For me, this connects directly to teaching and learning design. Not every task should be offloaded. Not every use of AI supports growth. Some uses may help faculty save time or generate ideas. Others may unintentionally reduce the very struggle, reflection, and practice students need in order to learn. The real challenge is being intentional: - Where can AI support brainstorming, feedback, or efficiency? - Where should human judgment remain central? - And how do we help students understand that difference? I love ideas like this because they move the conversation beyond hype or fear and toward more thoughtful decisions about learning, teaching, and academic work. #AIinEducation #HigherEd #TeachingWithAI #InstructionalDesign #AIliteracy #EdTech

We hand people AI tools before we ever ask them what they know about themselves as potential users. As a researcher in performance psychology, and a former college writing instructor, that gap concerns me. At the start of every semester, before my students wrote a single word, I asked them to learn something about themselves. I administered a VARK assessment: a tool that identifies how a person best absorbs and processes information. Most of them had never been asked that question before. Not in school. Not anywhere. In Self-Determination Theory, autonomy doesn't mean doing whatever you want. It means being the origin of your own choices. And you cannot be the origin of your choices if you don't yet know yourself. This is precisely where AI use breaks down before it even begins. When a person sits down without knowing what they want to say, there is a vacuum. An emptiness waiting to be filled by their own thinking. The tool fills it instead. And a filled vacuum feels like help. But it is the quiet erosion of self-knowledge that makes authentic expression possible. Learning how to use AI effectively requires knowing yourself first. I've also seen the opposite: what happens when a person arrives knowing something about themselves and is given the room to use it. That comes next. So, I'm curious. Before you open the tool itself: do you know who you are without it? Next up: what a fisherman taught me about competence and why the struggle was never the problem. A note on process: I collaborated with AI when drafting this post. I knew myself before I opened the tool: scholar, GenXer, researcher, former writing teacher. The ideas, the experiences, the research, and the choice of lens through which to see it are all mine. AI helped me organize what was already there and articulate who I wanted to present in this post. (Former writing teacher. The irony writes itself...and so, eventually, did I.)

🎯 5 AI Tools for Students & Learning Study smarter, not harder. 1️⃣ Grammarly – Better writing 2️⃣ QuillBot – Rewrite content 3️⃣ Khanmigo – Personal tutor 4️⃣ Speechify – Listen to notes 🎧 5️⃣ Notion AI – Smart study planner 💡 AI = your personal teacher. Which tool would you use? 👇 #Students #Learning #AI

I once asked AI to “help me study”… It gave me a explanation, no examples, no questions… basically a digital textbook 🤦♂️ Then I realized — I didn’t need more information I needed the AI to behave like a teacher. That’s when these concepts finally clicked 👇 AI concepts simplified (Prompt vs Instructions vs Agent vs Skills) Prompt → What you ask (one-time request) 👉 Example: “Explain Newton’s laws in simple words” Instructions → How AI should behave (ongoing rules) 👉 Example: “Always explain like I’m a beginner” Agent → AI configured for a specific job (brain + purpose + tools) 👉 Example: A “Study Coach” that teaches, quizzes, and tracks progress Skills → Extra abilities/tools the agent can use (APIs, file reading, etc.) 👉 Example: Reading a PDF, checking weather, running code Model (GPT/Claude/Gemini) → The core intelligence (brain) Default setup → Hidden system configuration (acts like a basic agent) Custom Agent → Your own tailored AI with specific behavior + skills 💡 Key idea Agent chooses and uses skills based on your request 👉 Ask “Summarize this PDF” → Agent uses file-reading skill In simple terms: The model is the brain 🧠 The agent is how that brain is set up to act Skills are the tools it uses And prompts + instructions? That’s how you tell it what to do and how to do it — like a chef using different tools depending on the dish you order 🍳 #AI #ArtificialIntelligence #GenerativeAI #AIAgents #PromptEngineering #AIForBeginners #LearnAI #TechSimplified #AIConcepts #FutureOfWork

Stop asking AI to do your work. Start asking AI to teach you how to do the work yourself. One approach builds dependency. The other builds a $3K-$10K/month business. The complete framework (backed by Harvard, Stanford & MIT research): https://lnkd.in/g9shAPFx

You all may have noticed how often we reach for AI without really thinking about it. At atomcamp, AI is already embedded in my workflow. While designing a course, refining an assignment, or even just trying to untangle an idea, it has become a natural part of how I think. Somewhere along the way, it ceased to be merely a tool and began to feel more like a thinking partner. Recently, Anthropic Academy has introduced some online courses. As someone responsible for creating and designing recorded online courses at my company, I thought it would be a good idea to explore the "AI Fluency for Educators" course. What stood out wasn’t that it taught me something entirely new; it provided me with language for things I’ve been doing instinctively. But the course made me pause and ask a slightly uncomfortable question: Am I being intentional about it, or just efficient? The 4D framework: Delegation, Description, Discernment, Diligence — sounds simple, but it forces you to confront your own habits: - Am I 'delegating' things I should actually be thinking through myself? - Am I giving enough 'context', or just expecting good outputs magically? - Am I critically 'engaging' with responses, or just accepting what sounds right? - Am I being 'thoughtful' about where and how I’m using AI? Working in learning design, this hit differently. Because the way I use AI inevitably shapes how others learn to use it too. And I think that’s the shift I’m sitting with, not “how can AI make my work faster?” but “what kind of thinking is my workflow reinforcing?” Still figuring that out..

I just published my first Substack post. It is not a hot take about AI. It is not another thread about agents replacing everything. It is a simple introduction to how I think about this moment, and why I decided to start writing in public. Over the past few years, my work has moved deeper into AI systems: LLM applications, retrieval pipelines, and agentic workflows built for real-world use. That experience keeps bringing me back to the same questions: What actually works? What breaks in deployment? What does it take to build AI systems people can trust? That is the focus of this Substack. I will be writing about AI and agentic workflows as they show up in everyday work, software, commerce, and education, especially the gap between a convincing prototype and a system that is genuinely useful in practice. Education is a big part of why I care. AI can make learning more accessible and more adaptive, but only if we build it with care. That means thinking seriously about trust, feedback quality, and how to make learning genuinely better, not just easier. My goal is simple: not hype, not doom, but grounded writing about what is broken, what is promising, and how we build better from here. I would love to hear from others thinking about the same questions. What do you think matters most as AI moves from demos into real deployment?