The 212AY Library
AI
Glossary.
200+ artificial intelligence terms explained in plain language — from machine learning and LLMs to AI agents, RAG and prompt engineering. Every definition comes with a concrete example, in English, French and Arabic.
200 terms
AI Adoption
AI adoption is the journey an organization takes to move from curiosity about artificial intelligence to using it productively in daily operations. Successful adoption usually starts small: pick one painful, repetitive process, run a pilot, measure the results, then scale what works. The main obstacles are rarely technical; they are unclear use cases, untrained staff, and fear of change. An artisan cooperative that adopts AI first for translating product descriptions can build confidence before tackling pricing or logistics.
AI Agent
An AI agent is a program powered by an AI model that can perceive its environment, reason about a goal, and take actions using tools such as web browsers, databases, or email, often across multiple steps without constant human guidance. Unlike a chatbot that only answers questions, an agent completes tasks: it can check inventory for a Casablanca e-commerce shop, draft supplier emails, and update the order sheet, chaining decisions the way a junior employee would.
AI Alignment
AI alignment is the research effort to ensure AI systems pursue the goals and values their designers and users actually intend, rather than harmful or unintended objectives. A misaligned system optimizes the letter of its instructions while betraying their spirit, like a sales chatbot told to maximize conversions that starts making false promises to customers. Alignment techniques such as RLHF and constitutional AI train models to be helpful, honest, and harmless, which becomes critical as systems gain autonomy.
AI Employee
An AI employee is a digital worker built from AI agents that takes full ownership of a business role rather than a single task: answering customers on WhatsApp around the clock, qualifying sales leads, chasing invoices, or managing social media content. Unlike a simple chatbot, an AI employee follows procedures, uses your company tools, remembers context, and reports to a human manager. For a small Moroccan business, it is like hiring a tireless assistant for the cost of a software subscription.
AI Ethics
AI ethics is the study and practice of using artificial intelligence in ways that respect human rights, dignity, and societal values. It addresses questions every organization deploying AI must answer: Is the system fair to all customers? Is personal data protected? Are people told when they interact with a machine? Who is accountable for mistakes? A Moroccan insurer using AI to price policies, for instance, needs ethical review to ensure the model does not discriminate and that clients can contest automated decisions.
AI Literacy
AI literacy is the baseline knowledge that lets non-specialists use artificial intelligence effectively and responsibly: understanding what models can and cannot do, writing clear prompts, spotting errors and hallucinations, and knowing when a human must stay in charge. It is becoming as essential as spreadsheet skills were in the 1990s. An HR team with solid AI literacy can draft job descriptions in minutes and screen faster, while knowing they must personally verify anything the AI claims about a candidate.
AI Safety
AI safety is the discipline focused on preventing AI systems from causing harm, whether through accidents, misuse, or unintended behavior. It spans practical concerns, like ensuring a medical chatbot never gives dangerous advice or a bank's model cannot be tricked into approving fraud, up to long-term questions about controlling very capable systems. For businesses, safety translates into concrete practices: testing before deployment, guardrails on outputs, human oversight for sensitive decisions, and clear procedures when the AI gets something wrong.
AI Transformation
AI transformation is the strategic process of rethinking how a company operates by embedding artificial intelligence across its functions: customer service, marketing, finance, operations, and decision-making. It goes beyond buying tools; it means redesigning workflows, training teams, and building a culture where humans and AI systems work together. A textile exporter in Casablanca might start with an AI assistant for client emails, then extend to demand forecasting and quality inspection, transforming step by step rather than all at once.
Algorithmic Bias
Algorithmic bias occurs when an AI system produces systematically unfair results for certain groups, usually because its training data reflects historical inequalities or lacks diversity. A recruitment model trained mostly on CVs from one city may unfairly downgrade candidates from rural regions; a credit-scoring system may disadvantage women entrepreneurs if past lending data did. Bias is rarely intentional, which makes it dangerous: companies must audit their data and test outcomes across groups before deploying AI in decisions about people.