AI Domains

AI can analyse learning performance and suggest practice, explanations, or content suited to a student’s progress and needs. This supports personalised learning rather than treating every learner identically. Teaching everyone at one pace, ignoring needs, or blocking feedback removes the adaptation and support that make the educational application useful.

Cameras and sensors provide information about roads, signs, vehicles, pedestrians, and obstacles. AI can analyse that information to support perception, route planning, and driving decisions such as stopping or turning. Seat colour, tyre decoration, and window-glass weight are not the intelligent navigation tasks described in this example.

Pattern recognition is the process of finding repeated features, relationships, or regularities in data. AI can use such patterns to classify items, detect unusual cases, or make predictions. Merely storing data does not reveal its structure, and writing instructions or reducing energy use describes different activities rather than pattern recognition.

Prediction uses patterns learned from historical or current data to estimate a likely future event or value. For example, a model may estimate demand or rainfall from relevant past observations. Storage alone does not create a prediction, and resizing images or repeating fixed instructions does not infer a future possibility.

An AI system can combine weather, soil, crop, and historical data, identify patterns, and suggest choices such as a suitable crop, irrigation time, or sowing period. The farmer should still use local knowledge and judgement. Placing a chair, folding paper, or hanging a clock is a physical task, not data-based decision support.

A prediction model learns relationships or patterns from relevant examples and uses them to estimate an outcome for new input. For instance, past demand may help estimate future demand. The estimate can be uncertain and may fail when conditions change, but it is not based merely on luck, computer size, or random decisions without data.

A recommendation system uses information such as interests, previous choices, learning activity, or stated needs to rank and suggest relevant options. Recommending a suitable lesson from a student’s interests and earlier selections is therefore personalised recommendation. Giving everyone the same lesson or choosing randomly ignores useful information, while permanently deleting less-used lessons is content management, not recommendation.

When messages are assigned known labels such as spam, important, or general, the task is classification: an input is placed into one of predefined categories. Clustering can group similar messages without previously supplied labels, so it is a possible different method in some systems but is not the clearest description here. Regression predicts numerical values, while visualization presents data graphically.

An educational AI system can examine performance, responses, and progress to suggest practice that matches a learner’s current needs. This is called personalised or adaptive learning and may provide extra support or greater challenge. Giving every student identical questions ignores differences in learning, while limiting textbooks or preventing learning is not an educational benefit.

AI can assist trained healthcare professionals by analysing medical images and highlighting patterns that may indicate a disease. Such a system supports screening or clinical review; it does not replace professional judgment, patient consent, privacy protection, or further testing. Publicising private data, accepting every output blindly, or spreading incorrect advice would be unsafe and unethical.

Automatically examining message content and placing messages into suitable categories uses patterns or features to make a decision without a person sorting each item. That combines AI-based classification with automation. Manual sorting is not AI automation, while sending everything to one group or deleting everything ignores content and does not demonstrate meaningful intelligent classification.

Classification assigns an item to one of predefined labels using its features, such as deciding whether an email is spam or not spam. Clustering groups similar items without fixed labels, so it differs from classification. Predicting a numerical value is regression, while removing irrelevant information is data cleaning rather than classification.

A recommendation system can analyse a user’s previous searches, views, and purchases to estimate which products may be relevant and display personalised suggestions. Showing the same fixed list to everyone is not personalised, while displaying a logo or manually entering product names does not require AI-based user modelling. Recommendations should still be used with privacy awareness.

An AI education system can examine responses, progress, and common errors to recommend practice suited to a learner’s current needs. This may support adaptive or personalised learning, while teachers remain important for explanation, encouragement, and judgement. Giving identical work to everyone ignores differences, and replacing teacher communication or reducing resources does not improve learning.

Weather forecasts, soil moisture and nutrients, crop condition, and related field data can help an AI system identify patterns and support decisions about irrigation, fertiliser, disease risk, or crop management. These suggestions should be interpreted with local knowledge and expert judgement. The number of chairs, wall colour, or booklet size is unrelated unless deliberately used for another task.

The system can study signals such as songs played, skipped, replayed, or liked and detect patterns in a user’s interests. It then uses those patterns to personalise later recommendations. A fixed list, alphabetical selection, or random choice does not adapt to individual behaviour, so none of those alternatives explains improvement based on preferences.

Decision support means that AI analyses relevant data and provides summaries, patterns, predictions, or recommendations to help a person make an informed choice. It does not necessarily make or implement the final decision itself. Hiding options, changing decisions without information, or applying every result without human review describes poor or unsafe practice, not decision support.

A spam classifier can learn patterns from examples, such as wording, sender behaviour, links, or message structure, and apply them to new emails. It then predicts whether a message is spam or not. A calculator following predetermined arithmetic steps and a manually operated switch perform fixed actions; merely reading rules without analysing data is not this kind of AI classification.

The governing concept is a complete pictograph key or legend. If one full symbol equals 10 students, a half symbol may represent 5 students, but that interpretation should be stated rather than assumed. Without the partial-symbol value, readers may calculate different totals or compare categories incorrectly. The issue is not colour, decoration, or image size; it is ambiguity in the data explanation. Thus option C is correct.