Inlay

## Background Mastodon’s timeline doesn’t rely on dopamine-driven algorithms – it is chronological and consent-based, showing only posts from accounts you (the person using Mastodon) have followed. This focus on privacy and conscious consumption is what leads many people to join the Fediverse in the first place. It also places an unfair ultimatum on incoming users: You’ll have to make an effort to figure out who to follow, or your timeline will mostly be empty. Bluesky pioneered a brilliant solution to this “empty feed problem” in 2024, with the introduction of “Starter Packs”, a feature that allows users to curate and share their own collections of recommended accounts. We believe that these kinds of user-generated, curated collections could help people to find their tribe more quickly when they join the Fediverse. At the same time, envisioning a similar feature that prioritises user consent, _and_ works across a constellation of independent servers, is no small feat. In this blog post, we want to talk about bringing a similar concept to Mastodon and the Fediverse. We’ll use the word “Packs” to refer to the shareable collections of identifiers throughout, but we’ve not yet landed on final terminology - so, consider this word a placeholder, and not what this will definitely be called in Mastodon. ## Challenges and considerations We know that there have been existing efforts to make it easier to discover curated collections of users (for example, fedidevs.com offers “Starter Packs”). We’ve been happy to see these being shared, as they can help people discover interesting individuals and organisations to follow. We’d be equally happy to have the creators of these tools provide feedback on our own ideas 🙂 We believe that there are several ways to improve on the existing examples, that are more aligned with the values we try to bring to the Fediverse, and that offer more to the decentralised ecosystem as a whole. Firstly, it’s important to us that users have control over whether they appear in Packs on Mastodon. Early design explorations with our concept of Packs led us to the following possibilities: * Packs will become an extension of discovery. Users who wish to opt out entirely from Packs will be able to do so by disabling the existing setting, labelled _“Feature profile and posts in discovery algorithms”_. This will signal that an account cannot be added to a Pack. * Users will be notified when they are included in a Pack. Unlike on Bluesky, where users wishing to remove themselves from a Starter Pack must either report the Starter Pack, or block the user, users on Mastodon will have a more neutral mechanism to remove themselves from a Pack they do not wish to be part of. (note: we implemented something similar with the Quote Posts feature, where an original post can be removed from a quote post; this same idea would flow through to Packs). As always, federation presents its own challenges. Just as Mastodon users can follow people on other Fediverse servers, our goal is for them to also be able to find and interact with Packs created elsewhere in the Fediverse. When Alice creates a Pack on her server `example.com`, how does Bob on `example.online` get to know about it, and come to interact with it? What if `example.com` and `example.online` run different ActivityPub-compatible software? These questions can be addressed via established Fediverse discussion processes. ## Next steps We’re in the process of collaborating with other Fediverse developers on a Fediverse Enhancement Proposal (FEP), so that a common implementation for Packs can be made available to developers of any ActivityPub software. The initial work is now available on GitHub. The FEP process will be the place to direct any technical questions. Meanwhile, we’re also conducting broader research to understand overarching themes related to user onboarding, and how we can make things easier for people to get started on Mastodon. We expect to release an initial version of Packs, plus other minor improvements to onboarding, in Mastodon 4.6. In the meantime, the next stable release (Mastodon 4.5) is right around the corner! ### We want to hear your thoughts We want to make Packs a great feature for discovery and onboarding! If you have thoughts on the ideas described above (beyond the technical aspects that will be part of the FEP), contact us at [email protected]. We may not be able to respond to every individual message, but we’ll be reading every piece of feedback to learn more about your ideas.

Additional content now available: Complete Print Edition (ISBN: 9798249845469) is available. For faculty who prefer distraction-free reading, marginalia, and a copy they can pass to a colleague, the print format is designed as a physical Cognitive Gym. It includes three additional chapters and a Workbook & Action Guide not in this OER: Chapter 5: Ogres Have Layers - Why "AI in Education" Is Four Things, Not One. Proposes four distinct layers of AI integration (AI Literacy, AI for Edu, AI in Edu, AI of the Profession), each with different governance structures, audiences, and evaluation criteria. Introduces the Displacement Clock for calibrating urgency across professional functions. Chapter 6: Jobs & The New Frontier. Counters the scarcity mindset with nine documented engines of AI job creation, from occupational decomposition to Jevons Paradox to the emerging compliance economy, showing how each engine implies different student preparation. Connects every guidebook framework to the abundance classroom. Chapter 7: AI Companions and the Boundaries of Care. Maps the companion ecosystem from study tools through emotional support to intimate attachment. Addresses why students turn to AI for emotional support, when that becomes harmful, and what faculty can do without becoming therapists. Engages the philosophical question of AI moral status through David Gunkel's relational ethics framework. Workbook & Action Guide: Twelve hands-on activities across four sessions, designed for workshops, reading groups, or individual professional development.Chapters 1–4 remain free and open access under CC BY 4.0. Find more on the companion website: dataii.com/ai/guidebook Introductory video. Artificial intelligence is affecting higher education, but not always in the ways we hoped. MIT neuroscience research shows that heavy AI reliance weakens neural connectivity and diminishes independent reasoning capacity, a process known as cognitive atrophy. Students who outsource their thinking to AI graduate with credentials but without the cognitive competence those credentials are supposed to represent. Meanwhile, faculty face a parallel challenge: how to use AI productively for their own work without surrendering the intellectual engagement that makes teaching meaningful. This guidebook offers a third path between banning AI and surrendering to it. Rather than treating AI as either a threat to be resisted or a shortcut to be embraced, it provides practical, research-grounded frameworks for using AI to strengthen thinking, both the educator's and the student's. The central argument is that AI should function as a cognitive gym, not a cognitive elevator: a tool that adds productive friction and challenge rather than removing it. The guidebook is organized around four core frameworks, each addressing a different dimension of AI integration in higher education: Cognitive Triage helps educators reclaim time by distinguishing between work worth delegating to AI (FLUFF: Formatting, Layouts, Under-the-hood, Filing, Filtering) and ideas worth protecting for human thought (SPARK: Specific, Persuasive, Authentic, Rigorous, Keen-insight). This framework uses a harvesting-vs.-seeding metaphor to help faculty identify where speed and automation are appropriate (transactional tasks with capped payoffs) and where struggle and investment produce lasting value (growth-oriented work with uncapped payoffs). The Intelligent Gearbox reframes AI prompting as a pedagogical skill. Understanding that AI is a probability engine, not a calculator, changes how educators interact with it. The chapter presents a four-gear progression of prompting techniques (One-Shot, Few-Shot, Chain of Thought, and Agentic) and reveals the guidebook's most powerful insight: the same principles that produce better AI outputs also produce better student learning. Scaffolding is not spoon-feeding; it is good instructional design. Every time you improve your prompts, you are practicing the skills that improve your teaching. The Cognitive Gym reverses the lens from faculty efficiency to student development. When it comes to learning, the goal is not to remove friction but to add it strategically. This chapter introduces Progressive Overload (using AI as a coaching partner that increases challenge), a five-step AI Audit verification protocol (Assumptions, Sources, Counter-Evidence, Auditing, Cross-Model) that shifts assessment from content generation to verification, and the VINE Framework (Vivid, Insightful, Narrative, Evident) for developing editorial taste, the judgment that distinguishes average from excellent, which AI cannot replicate. Analog Checkpoints provide verification tools for confirming genuine cognitive engagement. The Intelligent Simpleton addresses the most overlooked barrier to AI adoption: professional identity. The greatest obstacle to learning is not ignorance but ego, the need to appear as a know-it-all. Drawing on neuroplasticity research showing that growth happens at the edge of ability, this chapter explores how to use AI as a judgment-free zone for asking basic questions, how to overcome authenticity and institutional barriers, and why the courage to appear as a beginner is the path to remaining an expert. The guidebook is designed for practical application, not passive reading. Each chapter includes conceptual frameworks, concrete examples spanning both academic and technical career disciplines, actionable strategies that can be implemented immediately, and reflective prompts for deeper engagement. It validates educators' legitimate concerns about AI's impact on learning before offering solutions, building credibility through shared understanding of the challenges rather than dismissing skepticism. A companion set of faculty worksheets (published separately on Zenodo) provides structured activities for each chapter, designed for use in workshops, learning communities, or self-guided professional development. Companion site for updates and resources: dataii.com/ai/guidebook Table of Contents The Learn-It-All Educator — Introduction: As a Human Thinketh, Our Learning Mindset The Trap The Reframe What This Guidebook Offers How to Use This Guidebook Limitations and What This Guidebook Does Not Cover Chapter 1: Cognitive Triage — Managing Educator Workload in the Age of AI 1.1 Harvesting vs. Seeding: Two Types of Academic Work 1.2 FLUFF: The Work Worth Delegating 1.3 SPARK: Ideas Worth Thinking Putting It Together: From FLUFF to SPARK Chapter 2: The Intelligent Gearbox — Advanced Prompting as Pedagogical Skill 2.1 AI Is a Probability Engine 2.2 Shifting Through the Gears 2.3 From Zero-Shot Prompting to Zero-Shot Teaching Putting It Together Chapter 3: The Cognitive Gym — Pedagogy and Student Assessment in the Age of AI 3.1 Progressive Overload and the Review Board 3.2 The Verification Protocol (Academic Integrity) 3.3 The VINE Framework for Taste 3.4 The Analog Checkpoint: When Performance Mimics Engagement Putting It Together Chapter 4: The Intelligent Simpleton — Professional Mindset for the Age of AI 4.1 The Ego Trap 4.2 Beyond Ego: The Authenticity and Institutional Barriers 4.3 Embracing the Learn-It-All Culture 4.4 The Courage to Play the Simpleton References Appendix: Companion Worksheets for Learn-It-All Educator