Definition
AI stands for Artificial Intelligence — systems built to perform tasks that would normally require human intelligence.
AI is used in daily life in many everyday tools: voice assistants that respond to spoken commands, recommendation systems on shopping/streaming apps, spam filters in email, and navigation apps that suggest the fastest route.
Definition
Natural Language Processing (NLP)'s role is to let computers understand, interpret, and generate human language (spoken or written) — powering things like translation tools, chatbots, and voice assistants that need to make sense of what a person actually says or types.
An example of NLP in use: a translation app converting text from one language to another, or a chatbot understanding a typed customer question and responding appropriately.
Definition
The purpose of speech recognition technology is to convert spoken audio into text (or directly into commands a system can act on) — this is the technology that lets a voice assistant understand what you've said out loud.
A tool that uses AI to understand human voice is a voice assistant (like Siri, Google Assistant, or Alexa).
Definition
An example of robotics in everyday life: robotic vacuum cleaners that navigate and clean a home automatically.
The role of robots in factories is to perform repetitive, precise, or physically demanding manufacturing tasks (like assembling parts or welding) faster, more consistently, and often more safely than manual human labour for those specific repetitive tasks.
Definition
Recommendation systems work by analyzing a user's past behaviour (what they've watched, bought, or clicked on) and comparing it with patterns from many other users, to predict and suggest items that particular user is likely to be interested in next.
Definition
AI needs large and good-quality data because ML systems learn their behaviour directly from the data they're trained on — insufficient data leaves too few examples to learn reliable patterns from, while poor-quality (inaccurate, inconsistent, or unrepresentative) data teaches the system incorrect or misleading patterns, no matter how much of it there is.
The quantity of data refers to how much data is available for training (more examples generally help a model learn more robust, generalizable patterns), while the quality of data refers to how accurate, clean, relevant, and representative that data is — a large amount of poor-quality data can still produce a poorly-performing or biased AI system, since quantity alone can't fix fundamentally flawed data.
Automatic data collection is gathering data without needing constant manual human input, often through sensors or software running continuously — e.g. a fitness tracker automatically recording a wearer's steps and heart rate throughout the day.
Definition
Autonomous vehicles use tools such as cameras, radar, and LiDAR sensors (combined with AI image-recognition software) to detect traffic lights, road signs, and pedestrians in real time.
Two challenges of autonomous vehicles: correctly handling unpredictable real-world situations (like sudden pedestrian movement or unusual weather conditions), and the significant safety/legal responsibility involved if the vehicle's AI makes an incorrect decision.
Definition
Bias in an AI system is when its outputs systematically and unfairly favour or disadvantage certain groups, usually because the data it learned from itself contained that same imbalance.
Fairness is important in AI because a widely-used AI system's decisions (in hiring, lending, or law enforcement, for example) can significantly affect real people's lives — an unfair, biased system can cause real, repeated harm at scale, which is exactly why AI systems should be built and checked to treat different groups of people equitably.
AI systems should be transparent (their decision-making process should be explainable and understandable, not a complete 'black box') so that people affected by an AI decision can understand why it was made, and so problems like bias can actually be identified and corrected.
The difference between data security and privacy in AI systems: data security is about protecting data from unauthorized access, theft, or misuse (keeping it safe from outside threats); privacy is about respecting individuals' rights over their own personal data — what's collected about them, how it's used, and who it's shared with (even among those who are authorized to access it).
Surveillance is the use of AI-powered systems (like facial recognition or activity tracking) to monitor people's behaviour or movements; it must be used carefully because, while it can improve security, excessive or unchecked surveillance can seriously violate personal privacy and civil liberties if not properly limited and regulated.
Remember
AI designers have a big responsibility for how their systems behave because the choices they make in gathering training data, designing algorithms, and setting the system's goals directly shape every decision the AI later makes at scale — a flaw introduced at the design stage doesn't just affect one decision, it can repeat itself across millions of automated decisions the system makes afterward.
Definition
A challenge of using AI is that it can behave unpredictably or make mistakes when faced with data or situations very different from what it was trained on, and it can inherit and amplify biases present in its training data if these aren't carefully checked for.
Two new job roles created because of AI: AI/ML engineers (who build and train AI models) and AI ethics specialists (who assess and manage the fairness, safety, and ethical impact of AI systems).
Quick Recap
- ●AI Artificial Intelligence, used daily via voice assistants, recommendations, spam filters, navigation
- ●NLP: computers understanding/generating human language (translation, chatbots). Speech recognition: converts spoken audio to text/commands
- ●Robotics: repetitive/precise physical tasks (factories, home robots). Recommendation systems: predict interests from past behaviour patterns
- ●Data quantity (how much) vs quality (how accurate/clean/representative) — both matter; poor quality data harms AI regardless of amount
- ●Autonomous vehicles: cameras/radar/LiDAR AI to detect surroundings; challenges include unpredictable situations and safety responsibility
- ●Ethics: bias (unfair systematic favouritism from skewed data), fairness (equitable treatment), transparency (explainable decisions)
- ●Data security (protection from misuse/theft) vs privacy (control over one's own personal data); surveillance needs careful limits
- ●AI designers bear major responsibility, since design-stage flaws repeat across every automated decision the system makes
