ICDL AI Essentials flashcards
ICDL AI Essentials · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- Artificial intelligence
- Artificial intelligence is the field of building computer systems that perform tasks associated with human perception, reasoning, learning, language, or decision-making.
- Machine learning
- Machine learning is a subset of AI in which systems learn patterns from data and improve with experience rather than following only hand-coded rules.
- Training
- Training is the process of adjusting a model's internal parameters using data so it can make predictions or generate outputs for a task.
- Inference
- Inference is the stage where a trained model applies learned patterns to new input to produce a prediction, classification, or generated response.
- Supervised learning
- Supervised learning trains a model on labeled examples, pairing inputs with correct outputs so it can predict labels for new, unseen inputs.
- Generative AI
- Generative AI creates new text, images, audio, or code by learning statistical patterns in training data and producing plausible outputs.
- Narrow AI
- Narrow AI is designed for a specific task or limited set of tasks, such as translation or recommendation, rather than general human-like intelligence.
- Neural network
- A neural network is a computing model made of layered, weighted connections that learns patterns by adjusting weights during training.
- Recommendation system
- A recommendation system predicts which products, videos, or posts a user may prefer by analyzing behavior patterns and similarities to other users or items.
- Voice assistant
- A voice assistant uses speech recognition and natural language processing to interpret spoken requests and respond or control connected devices.
- Spam filter
- A spam filter classifies email or messages as unwanted by detecting patterns such as suspicious senders, keywords, links, or abnormal behavior.
- Fraud detection
- Fraud detection uses AI to flag transactions or account activity that deviates from expected patterns and may indicate theft or abuse.
- Machine translation
- Machine translation automatically converts text or speech from one language to another by modeling patterns learned from multilingual data.
- Route optimization
- Route optimization uses traffic, distance, and timing data to predict travel conditions and suggest efficient paths for navigation or delivery.
- Chatbot
- A chatbot is a conversational AI interface that interprets user messages and supplies automated replies, often handling routine service requests.
- Smart home automation
- Smart home automation uses connected sensors and AI rules to control lighting, heating, security, or appliances based on presence, schedules, or voice commands.
- Prompt
- A prompt is the instruction or input given to an AI system that specifies the task, context, constraints, and desired output format.
- Few-shot prompting
- Few-shot prompting gives a model a small number of example input-output pairs so it can infer the desired pattern or style.
- Hallucination
- A hallucination is an AI output that is fluent and confident but factually wrong or unsupported by reliable evidence.
- Prompt iteration
- Prompt iteration is the repeated process of testing an AI output, refining instructions or examples, and re-running the request to improve results.
- System instruction
- A system instruction is a high-level directive that sets an AI assistant's role, boundaries, or response style before user requests are processed.
- Human review
- Human review is the checkpoint where a person checks AI-generated work for accuracy, safety, fairness, and fitness before it is used or published.
- Temperature setting
- A temperature setting controls randomness in some AI models; lower values produce more predictable text and higher values produce more varied text.
- Output format
- An output format specifies the structure a user requests from an AI tool, such as a table, list, JSON, or short summary.
- Algorithmic bias
- Algorithmic bias is systematic unfairness in AI outputs that can arise from biased training data, flawed labels, or design choices that disadvantage groups.
- Transparency
- Transparency in AI means making relevant information about a system's purpose, data use, limitations, and decision process available to affected people.
- Accountability
- Accountability means that a person or organization remains responsible for AI-assisted decisions and can explain and correct harmful outcomes.
- Informed consent
- Informed consent means people understand what data or AI processing is involved and voluntarily agree before participation or use.
- Copyright risk
- Copyright risk in AI involves using protected works for training or generating outputs that may reproduce, imitate, or improperly incorporate someone else's creative material.
- Data leakage
- Data leakage occurs when sensitive information entered into or processed by an AI tool is exposed to unauthorized people or systems.
- Prompt injection
- Prompt injection is an attack where hidden or malicious instructions in content cause an AI system to ignore its intended rules or reveal protected information.
- Deepfake
- A deepfake is synthetic image, audio, or video content generated or manipulated by AI to depict people or events that did not occur.