# AI-Empowered Youth: Think, Build, Lead (Builder, ages 14-17) > Coming soon. An online, self-paced AI course from ARFA, the TIBLOGICS AI Academy. Beginner level, about 23.5 hours, in English and French, with a verifiable certificate. Web page: https://tiblogics.com/learning-box/ai-empowered-youth-builder ยท All tracks: https://tiblogics.com/learning-box.md ## Key facts - Level: Beginner - Time: about 23.5 hours including hands-on work (12 modules, 57 lessons, 12 labs); self-paced, no deadline - Price: announced when the track opens - Certificate: AI-Empowered Youth: Builder, with a public verification page - Languages: English and French ## About You will graduate into a world where AI is part of every job, every business and every decision. Builder is for teenagers aged 14 to 17 who want to understand AI properly, think sharply about it and build real things with it, not just scroll past it. The program runs in four seasons. Season 1, Understand, takes the AI in your world apart: the recommendation engine behind your feed and what it is really optimising, how neural networks learn (with code you run and change on the page), why models fail and become unfair, and the engineering behind game AI, pathfinding, face unlock, voice assistants and on-device AI. Later seasons move from understanding to thinking critically with AI, building your own projects and tools, and leading: making the case for AI used fairly and well in your school, community or future business. Every lesson follows one rhythm: learn the real concept with an example from your world, play with it in an interactive tool or code playground, build something, and reflect. You will use the thinking tools engineers, scientists and founders rely on: first principles, the 5 Whys, systems maps and feedback loops, Goodhart's law, claim-evidence-reasoning and the debugging mindset. The labs are harder: audit and redesign a recommender, peer-review an AI project, program a game character that cannot be tricked. Everything happens on the platform. The practice AI runs on ARFA's own safe system, so there are no outside accounts to create and no personal details to share. Who it is for: Teenagers aged 14 to 17 who want to understand how AI works and build with it. No coding experience needed; code is introduced gently where it helps. Parents and carers can follow along. ## What you will be able to do - Explain the pipeline behind a recommendation feed and redesign its objective with Goodhart's law in mind - Describe how a neural network learns, including weights, loss and gradient descent, and train one in code - Diagnose model failures such as overfitting, shortcuts and distribution shift, and test for bias by group - Explain how game AI, pathfinding, face recognition and voice assistants work, and their trade-offs - Decide where an AI feature should run, on device or in the cloud, with a reasoned argument - Apply systems thinking, the 5 Whys and claim-evidence-reasoning to real problems - Write prompts as specs, test them against criteria and turn the best into reusable templates - Create stories, art, music, web apps and games with AI while respecting copyright, consent and disclosure - Design, red-team and safely automate a helper bot, with a human in the loop where it matters ## Curriculum ### Module 1: How your feed knows you Take apart the recommendation engine behind video and social apps: the pipeline, the signals, what it optimises and why that matters, the loops that build bubbles and rabbit holes, the persuasive design that keeps you scrolling, and an experiment to retrain your own feed. - Inside a recommendation engine (15 min) - What the feed is optimising, and Goodhart's law (14 min) - Filter bubbles, rabbit holes and feedback loops (15 min) - Persuasive design: why stopping is hard (13 min) - Run a feed experiment (16 min) - Module quiz ### Module 2: Inside the AI brain Learn how machine learning really works: features, labels and test sets; what a neuron computes and how gradient descent trains a network; why models overfit, take shortcuts and get fooled; how training data shapes fairness; and why language models are prediction, not minds. - From rules to learning: features, labels and models (14 min) - Neural networks from the inside (18 min) - How models fail: shortcuts, overfitting and adversarial tricks (16 min) - Training data is destiny: bias and fairness (15 min) - Language models: prediction, not a mind (15 min) - Module quiz ### Module 3: AI in your games and phone Look under the hood of everyday AI: state machines and game economies, pathfinding with breadth-first search and A, face detection and recognition, voice assistant and autocomplete pipelines, and the engineering trade-offs between running AI on your device and in the cloud. - Game AI: state machines, behaviour trees and game systems (15 min) - Pathfinding: how characters find the shortest route (16 min) - Camera AI: detection, recognition and face unlock (15 min) - Voice assistants and autocomplete: the prediction pipeline (14 min) - On-device or cloud: the engineering trade-offs (16 min) - Module quiz ### Module 4: Fake or real? Learn how synthetic images, voices and text are made and why seeing is no longer believing, why AI hallucinates, and how professional fact-checkers work: SIFT, lateral reading, tracing claims to the source, and building claim, evidence and reasoning that holds up before you share. - Seeing is no longer believing (14 min) - Deepfake forensics, and why tells expire (15 min) - Why fluent is not the same as true (15 min) - Think like a fact-checker: SIFT and lateral reading (16 min) - Claim, evidence, reasoning: arguments that hold (16 min) - Module quiz ### Module 5: Systems are everywhere Think like a systems analyst about the world you live in: a school canteen, a game economy, a city's traffic, a school timetable, a family business. Map stocks and flows, trace reinforcing and balancing loops, find bottlenecks and delays, predict unintended consequences, and choose leverage points that actually change outcomes. - Seeing systems: parts, stocks, flows and purpose (14 min) - Game economies: sources, sinks and runaway loops (15 min) - Feedback loops, delays and oscillation (15 min) - Bottlenecks and the timetable problem (15 min) - Leverage points and unintended consequences (16 min) - Module quiz ### Module 6: Your data, your power Understand what your data reveals, including what apps infer about you without being told, how the data business model works, and how to take control with permissions and settings. Then go inside AI bias: how unrepresentative data, proxies and feedback loops produce unfair systems, who gets left out, and what meaningful consent and your data rights look like. - Your data trail and what it reveals (14 min) - Permissions, settings and the business model (15 min) - Bias in, bias out (16 min) - Consent, fairness and your rights (15 min) - Module quiz ### Module 7: AI and your mind Look at the systems competing for your attention and how engagement optimisation shapes your feed and your mood; separate using AI to learn from outsourcing your thinking; understand what AI companions are designed to do and where their limits are; and design your own system for focus, learning and wellbeing. - The attention economy, from the inside (15 min) - Learning with AI vs outsourcing your thinking (16 min) - AI companions: what they are, and what they cannot be (15 min) - Design your own system for focus and wellbeing (16 min) - Module quiz ### Module 8: Prompt power Write prompts the way engineers write specs: a clear goal, the context only you know, constraints, examples and an output format. Then iterate one change at a time, test outputs against criteria you set in advance, and turn your best prompts into reusable templates. - A prompt is a spec: goal, context and constraints (14 min) - Examples and output formats (15 min) - Judge the output: criteria, tests and checks (16 min) - Iterate like an engineer (15 min) - Reusable prompts: templates, roles and versions (15 min) - Module quiz ### Module 9: Create: stories, art and music Make stories, images and music with AI while keeping the creative decisions yours: co-creation workflows, prompting for visuals and sound, copyright and licensing basics, attribution that other creators can rely on, honest AI disclosure, and the ethics of using real people's faces and voices. - Co-creation: you direct, AI assists (15 min) - Stories and images: prompting for creative work (16 min) - Music, voice and sound (15 min) - Copyright, licences and attribution (16 min) - Disclosure and deepfake ethics for creators (15 min) - Module quiz ### Module 10: Vibe coding 101 Build small web apps by describing them to an AI, and stay in control while you do: write a spec first, understand the three layers of a web page (HTML, CSS and JavaScript), read and question the code the AI gives you, debug methodically, and grow an app in small, tested, saved steps. - Vibe coding: build by describing, starting with a spec (15 min) - HTML, CSS and JavaScript: the three layers of a web app (16 min) - Read and question AI code (16 min) - Debugging: find it, understand it, fix it (16 min) - Small steps, tests and versions: level up your app (15 min) - Module quiz ### Module 11: Game studio Design small games the way studios do: a game design document, a core loop worth repeating, mechanics and feedback, numbers balanced so the game stays fair and interesting, levels that teach before they test, characters whose behaviour you control, and playtests that show you what players really do. - The game design document and the core loop (15 min) - Mechanics, feedback and balancing (16 min) - Level design: teach, test, twist (16 min) - NPC behaviour: rules, states and AI characters (16 min) - Playtesting and iteration (15 min) - Module quiz ### Module 12: Bots and agents Understand AI agents as loops with goals, tools and memory, then design, test and protect a helper bot of your own: its personality, rules and knowledge, red-teaming and prompt injection, and where a human must stay in the loop when you automate a real task. - What an agent is: goals, tools, memory and the loop (15 min) - Design a helper bot: personality, rules and knowledge (16 min) - Test, red-team and defend against prompt injection (17 min) - Human in the loop: automate a real task safely (17 min) - Module quiz ## How you are assessed - A quiz after each module (80% to pass, retakes allowed). - Each certificate has a public page at https://tiblogics.com/certificates/ that anyone can open to check it.