8 questions · about 4% of the marks we indexed · examined in 7 of 9 sessions.
Questions by year
2026
2026 May/June · Paper 12 · Q6 (b)4 marksArtificial intelligence, expert systems and machine learning
2025
2025 Oct/Nov · Paper 12 · Q7 (a, c)7 marksArtificial intelligence, expert systems and machine learning · Robotics
2025 May/June · Paper 12 · Q5 (a, b(i), b(ii), b(iii))11 marksArtificial intelligence, expert systems and machine learning · Robotics
2024
2024 Oct/Nov · Paper 12 · Q8 (b, c)6 marksArtificial intelligence, expert systems and machine learning
2024 May/June · Paper 12 · Q10 (a, b)7 marksArtificial intelligence, expert systems and machine learning
2023
2023 Oct/Nov · Paper 12 · Q10 (a, b)7 marksArtificial intelligence, expert systems and machine learning
2023 Oct/Nov · Paper 12 · Q11 (b, c)4 marksAutomated systems
2023 May/June · Paper 12 · Q9 (a, b)5 marksArtificial intelligence, expert systems and machine learning
No questions match this filter.
References only, not question text. Open Official paper ↗ where Cambridge publishes the paper,
or see the official Cambridge past-papers page; for other papers, ask your
school. “Paper 42” means Paper 4, variant 2.
Practice questions
Original questions written by HomiLabs in the style of the exam, with answers and mark schemes. They are not
Cambridge past paper questions and have not been reviewed by Cambridge. Try each one before opening the answer.
Q1Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 4 marks
(a)State what is meant by artificial intelligence (AI).[1]
(b)State three characteristics of AI.[3]
Show answer
(a) A branch of computer science dealing with the simulation of intelligent (human-like) behaviour by computers.
1 mark.
(b) Collection of data; rules for using the data; the ability to reason; (can include) the ability to learn and adapt.
1 mark for each of three characteristics.
Q2Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 4 marks
(a)An expert system helps farmers in Punjab identify diseases in wheat crops. State the four main components of an expert system.[4]
Q3Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 3 marks
(a)Give three examples of tasks in which expert systems are used.[3]
Show answer
(a) Any three, e.g. medical diagnosis; diagnosing faults in cars or machinery; identifying plant or crop diseases; mineral or oil prospecting; financial planning/credit decisions; chess playing; weather forecasting; tax advice.
1 mark for each example.
Q4Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 3 marks
(a)Describe what is meant by machine learning.[2]
(b)Give one example of a system that uses machine learning.[1]
Show answer
(a) Machine learning is a type of AI in which a program can automatically adapt/alter its own processes or data, learning from experience (training data or past results) without being explicitly reprogrammed.
1 mark: program changes its own rules/data. 1 mark: based on experience/data, without a programmer changing the code.
(b) Any one, e.g. an email spam filter; a video/music recommendation system; fraud detection in banking; face recognition; a predictive-text keyboard; a game opponent that improves.
1 mark.
Q5Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 4 marks
(a)Match each component of an expert system to its description.Components: knowledge base, rule base, inference engine, interfaceDescriptions:1. Allows the user to enter answers and see the results.2. Stores the facts about the subject area.3. Stores the IF … THEN rules used to reach conclusions.4. Applies the rules to the facts and the user's answers to reach a conclusion.[4]
Q6Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 3 marks
(a)A car-repair garage in Lahore uses an expert system to diagnose engine faults. Give two benefits to the garage of using the expert system.[2]
(b)Give one drawback of relying on the expert system.[1]
Show answer
(a) Any two, e.g. gives consistent diagnoses; can be faster than a human working through every possibility; less experienced mechanics can diagnose complex faults; the knowledge of many experts is available at once; gives a probability/confidence for the diagnosis; available at any time.
1 mark for each benefit.
(b) Any one, e.g. it is only as good as the data and rules entered; it cannot use common sense or notice unusual symptoms not in its rules; it is expensive to set up and maintain; mechanics may lose skills.
1 mark.
Q7Artificial intelligence, expert systems and machine learning · Core · Paper 1 style · 2 marks
(a)State the purpose of the inference engine in an expert system.[2]
Show answer
(a) It applies the rules in the rule base to the facts in the knowledge base (and the user's answers) to draw conclusions/reach a diagnosis; it decides which questions to ask next.
1 mark: uses rules from the rule base with the knowledge base facts. 1 mark: to reach a conclusion/decide the next question.
Q8Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 5 marks
(a)A clinic uses an expert system to suggest possible causes of a patient's symptoms. Describe how the expert system reaches its suggestion.[5]
Show answer
(a) The interface asks the user questions about the symptoms; the user's answers are entered through the interface; the inference engine compares the answers with the facts in the knowledge base; using the rules in the rule base (IF … THEN); it may ask further questions to narrow down the possibilities; it reaches a conclusion and the interface displays the possible causes, with probabilities and an explanation of the reasoning.
1 mark each for any five points in a logical order.
Q9Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)An online video service recommends new films to each viewer. Explain how machine learning allows the recommendations to improve over time.[4]
Show answer
(a) The system collects data on what each viewer watches, rates, skips or searches for; it finds patterns in this data and compares it with similar viewers; it uses the patterns to predict films the viewer will like; it learns from whether the viewer watches the recommended films and adjusts its own rules/weightings automatically, so later recommendations are more accurate.
1 mark each for any four points.
Q10Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)A geology company uses an expert system to decide where to drill for minerals. Explain why the knowledge base must be built with help from human experts.[2]
(b)Explain the difference between the knowledge base and the rule base.[2]
Show answer
(a) The knowledge base holds the facts/expertise of the subject area, such as the rock types linked to minerals; this knowledge comes from experienced geologists, so they must provide/check it for the conclusions to be accurate.
1 mark: knowledge base stores expert facts. 1 mark: these come from human experts / accuracy depends on them.
(b) The knowledge base stores facts about the subject (e.g. properties of rocks); the rule base stores the logical IF … THEN rules that link the facts to conclusions.
1 mark for each.
Q11Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)An email provider uses a spam filter that uses machine learning. Describe how the filter learns to identify spam emails.[3]
(b)Explain why a spam filter that uses fixed rules written by a programmer is less effective.[1]
Show answer
(a) It is trained with examples of emails labelled spam or not spam; it identifies features/patterns common in spam (words, links, senders); when users mark emails as spam or not spam, it uses this feedback to adjust its own rules, so its accuracy improves without being reprogrammed.
1 mark each for three points.
(b) Spammers keep changing their messages; fixed rules cannot adapt to new patterns unless a programmer updates them, whereas machine learning adapts automatically.
1 mark.
Q12Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)A chat assistant on a bank's website answers customers' questions. Explain how the assistant shows each of these characteristics of AI.(i) Collection of data(ii) Rules for using the data(iii) Ability to reason[3]
(b)Explain how the assistant could also show the ability to learn and adapt.[1]
Show answer
(a)(i) It stores data such as previous questions, answers and account information.(ii) It follows rules about how to match a question to an answer or action.(iii) It uses the rules and data to work out the most likely meaning of a new question and the best reply.
1 mark for each characteristic applied to the chat assistant.
(b) It can change its own responses based on feedback, e.g. whether customers found an answer helpful or rephrased the question, so it gives better answers in future.
1 mark.
Q13Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)A plant-identification app uses an expert system. A user answers questions about leaf shape, flower colour and height. Describe the role of the user interface in this system.[2]
(b)The app shows a percentage next to each possible plant. Explain why.[1]
(c)Suggest one reason why the app may give a wrong identification.[1]
Show answer
(a) It presents the questions to the user (often as multiple-choice/yes–no options) and accepts the answers; it displays the result (the identified plant), possibly with a probability and an explanation of the reasoning.
(b) The inference engine may not be certain; the percentage shows how likely/probable each match is, based on how well the answers fit the rules.
1 mark.
(c) Any one, e.g. the user gives inaccurate answers; the plant is not in the knowledge base; the rules are incomplete.
1 mark.
Q14Artificial intelligence, expert systems and machine learning · Standard · Paper 1 style · 4 marks
(a)Explain two differences between an expert system and a machine-learning system.[4]
Show answer
(a) Any two, e.g. an expert system's knowledge and rules are entered by humans/experts, whereas a machine-learning system develops its own rules from data; an expert system does not change unless someone edits its knowledge base or rules, whereas machine learning adapts automatically as it gets more data/experience; an expert system can explain its reasoning by showing the rules used, whereas a machine-learning system's reasoning is often harder to explain.
For each difference: 1 mark for the expert system point, 1 mark for the matching machine-learning point.
Q15Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 5 marks
(a)A team is building an expert system to help engineers diagnose faults in solar-power systems in rural Sindh. Describe the steps the team would follow to set up the expert system.[5]
Show answer
(a) Gather information/knowledge from experienced engineers (experts) and manuals; enter the facts into the knowledge base; write the IF … THEN rules into the rule base; create the inference engine (or use an existing shell) to apply the rules; design the user interface with questions and screens for the results; test the system using faults with known causes and compare the results with the experts' answers; refine the rules and knowledge base as necessary.
1 mark each for any five points in a sensible order.
Q16Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 5 marks
(a)A bank uses machine learning to detect fraudulent card payments. Explain how the system decides whether a new payment might be fraudulent, and how it improves over time.[4]
(b)Give one disadvantage to customers of this system.[1]
Show answer
(a) It is trained on large amounts of past payment data labelled as genuine or fraudulent; it learns the normal spending patterns of each customer (location, amount, time, type of shop); a new payment that does not fit these patterns is flagged/blocked or the customer is asked to confirm; when the outcome is known (customer confirms genuine or reports fraud), the system uses it to adjust its own rules, so fewer genuine payments are flagged and more fraud is caught.
1 mark each for any four points.
(b) Any one, e.g. genuine payments may be blocked (false positives), for example when travelling abroad; concern about personal data being analysed.
1 mark.
Q17Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 5 marks
(a)A medical expert system asks: 'Does the patient have a fever?' The answer is YES. The rule base contains:Rule 1: IF fever AND rash THEN condition PRule 2: IF fever AND cough AND NOT rash THEN condition QRule 3: IF NOT fever AND cough THEN condition RExplain how the inference engine uses these rules to decide which question to ask next, and how it reaches a conclusion if the next answers are 'rash: NO' and 'cough: YES'.[4]
(b)Explain why a doctor should still check the result.[1]
Show answer
(a) Fever is YES, so Rule 3 cannot apply and is ignored; Rules 1 and 2 are still possible, so the inference engine asks about a rash (or cough) next; rash is NO, so Rule 1 is ruled out; cough is YES, so all conditions of Rule 2 are true (fever, cough, no rash); the inference engine concludes condition Q and the interface displays it (with its reasoning).
1 mark: Rule 3 eliminated by the fever answer. 1 mark: next question chosen from the remaining rules. 1 mark: Rule 1 eliminated by rash = NO. 1 mark: Rule 2 satisfied → condition Q.
(b) The system can only use the rules and facts it has; it may miss symptoms or conditions not in its knowledge base, so a doctor's judgement is needed for safety.
1 mark.
Q18Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 6 marks
(a)A self-driving delivery van uses artificial intelligence. Explain how the van uses each of these characteristics of AI to drive safely through a busy city such as Karachi.(i) Collection of data(ii) Rules for using the data(iii) Ability to reason(iv) Ability to learn and adapt[4]
(b)Give two reasons why people might not trust the van.[2]
Show answer
(a)(i) Sensors and cameras continuously collect data about the road, traffic, pedestrians and signals.(ii) It has rules such as stop at a red light, keep a safe distance, give way to pedestrians.(iii) It combines the data and rules to decide what to do in a situation, e.g. predicts that a pedestrian will cross and slows down.(iv) It uses data from previous journeys (machine learning) to improve its decisions, e.g. learning how drivers behave at a particular junction.
1 mark for each characteristic applied to the van.
(b) Any two, e.g. it might react wrongly to an unusual situation it has not seen; sensors could fail in heavy rain or dust; it is unclear who is responsible after an accident; it could be hacked.
1 mark for each reason.
Q19Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 6 marks
(a)A university in Islamabad is choosing between an expert system and a machine-learning system to predict which students need extra support. Evaluate the two options, giving one advantage and one disadvantage of each.[4]
(b)Explain why the university must be careful about the data used to train a machine-learning system.[2]
Show answer
(a)Expert system – advantage: its rules are written by experienced tutors and it can explain why a student was flagged; disadvantage: it does not improve on its own and may miss new patterns unless the rules are updated.Machine learning – advantage: it can find patterns in large amounts of past student data and improves as more data is added; disadvantage: it needs a lot of good-quality (unbiased) data and its decisions can be hard to explain.
1 mark for each valid advantage/disadvantage (2 for each system).
(b) The system learns from the data it is given; if the data is incomplete or biased (e.g. from one group of students), it will learn biased rules and make unfair or inaccurate predictions.
1 mark: learns from training data. 1 mark: biased/poor data leads to unfair/inaccurate results.
Q20Artificial intelligence, expert systems and machine learning · Challenge · Paper 1 style · 5 marks
(a)A translation app uses machine learning. When it was first released it often translated Urdu phrases incorrectly. Explain how the app can become more accurate without programmers rewriting its code.[3]
(b)Explain why this is an example of AI but a basic calculator is not.[2]
Show answer
(a) It collects large amounts of example translations (and new text as people use it); users correct or rate translations, and this feedback is stored; the app analyses the data to find patterns and updates its own rules/model automatically, so future translations are better.
1 mark each for three points.
(b) The app shows the ability to learn and adapt / reason with data to produce a human-like result, changing its behaviour from experience; a calculator only follows fixed instructions and always produces the same result without learning.
1 mark: app learns/adapts or simulates intelligent behaviour. 1 mark: calculator follows fixed rules and does not change.
Q21Robotics · Core · Paper 1 style · 4 marks
(a)State the three main characteristics of a robot.[3]
(b)Give one example of a robot used in the home.[1]
Show answer
(a) It has a mechanical structure or framework; it has electrical components such as sensors, microprocessors and actuators; it is programmable.
1 mark for each characteristic.
(b) Any one, e.g. robot vacuum cleaner; robot lawn mower; window-cleaning robot; pool-cleaning robot.
1 mark.
Q22Robotics · Core · Paper 1 style · 4 marks
(a)Robots are used in many areas. Give one example of a task carried out by a robot in each of these areas.(i) Industry(ii) Agriculture(iii) Medicine(iv) Transport[4]
Show answer
(a) (i) e.g. welding or spray-painting car bodies; packing goods; moving items in a warehouse. (ii) e.g. picking fruit; weeding; milking cows; spraying crops with a drone. (iii) e.g. assisting surgeons in precise surgery; delivering medicines in a hospital; disinfecting rooms. (iv) e.g. driverless cars; autonomous trains; delivery drones.
1 mark for each suitable example.
Q23Robotics · Standard · Paper 1 style · 4 marks
(a)A car factory near Karachi uses robots to weld car bodies. Give two advantages to the factory of using robots for this task.[2]
(b)Give two disadvantages of using the robots.[2]
Show answer
(a) Any two, e.g. can work 24/7 without breaks; welds are consistent and accurate; can work in dangerous conditions (heat, fumes); faster than humans; lower long-term labour costs.
1 mark for each advantage.
(b) Any two, e.g. high initial cost to buy and set up; expensive maintenance; workers may lose their jobs or need retraining; robots cannot easily adapt to tasks they are not programmed for; production stops if a robot breaks down.
1 mark for each disadvantage.
Q24Robotics · Standard · Paper 1 style · 4 marks
(a)A robot lawn mower cuts the grass in a large garden. Describe the role of each of these components in the robot.(i) Sensors(ii) Microprocessor(iii) Actuators[3]
(b)Explain why the lawn mower is described as programmable.[1]
Show answer
(a)(i) Sensors collect data about the surroundings, e.g. detecting obstacles, the edge wire or rain.(ii) The microprocessor processes the sensor data and decides what to do according to its program, e.g. turn when an obstacle is detected.(iii) Actuators carry out the movement, e.g. turning the wheels and the cutting blade motors.
1 mark for each component.
(b) Its behaviour is controlled by a program that can be changed/reprogrammed, e.g. setting a new cutting schedule or area.
1 mark.
Q25Robotics · Standard · Paper 1 style · 4 marks
(a)A hospital uses a robotic arm to help surgeons with delicate operations. Explain two benefits of using the robotic arm.[4]
Show answer
(a) Any two explained, e.g. very precise and steady movement, so smaller cuts are made and patients recover faster; it does not get tired during long operations, so accuracy is maintained; it can be controlled remotely, so a specialist surgeon does not have to be present.
For each benefit: 1 mark for the point and 1 mark for the expansion.
Q26Robotics · Challenge · Paper 1 style · 6 marks
(a)A fruit farm in Swat is considering robots that pick apples. Explain how a picking robot could use sensors, a microprocessor and actuators to pick only ripe apples.[4]
(b)Discuss one advantage and one disadvantage of the robots for the farm workers in the area.[2]
Show answer
(a) Cameras/light sensors capture images of the apples and measure their colour/size; the microprocessor compares the readings with stored values for a ripe apple; if the apple is ripe, the microprocessor sends signals to the actuators; the actuators move the arm to the apple and the gripper picks it gently (using pressure sensors to avoid bruising); otherwise the robot moves on.
1 mark each for any four points.
(b) Advantage: e.g. workers no longer need to do tiring, repetitive work at heights, or new skilled jobs maintaining the robots. Disadvantage: e.g. fewer seasonal picking jobs, so loss of income for local workers.
1 mark for the advantage, 1 mark for the disadvantage.
Q27Robotics · Challenge · Paper 1 style · 5 marks
(a)A student says: 'A washing machine with a microprocessor is a robot.' Explain whether the student is correct by referring to the characteristics of a robot.[3]
(b)Explain how a robot can be described as having artificial intelligence.[2]
Show answer
(a) A washing machine has a mechanical structure (drum, motor) and electrical components (sensors, microprocessor, actuators such as motors and valves), but it is only programmable in the sense that the user chooses from fixed pre-set wash programs; it cannot be reprogrammed to carry out different tasks. So it is usually classed as an automated system (embedded system) rather than a robot, and the student is not correct.
1 mark: has mechanical structure and/or electrical components. 1 mark: only fixed pre-set programs / cannot be reprogrammed for other tasks. 1 mark: conclusion with reason (automated/embedded system, single dedicated task). Accept a reasoned answer either way.
(b) A robot has AI if it can collect data and make decisions/reason for itself rather than only follow fixed instructions, and can learn/adapt from experience, e.g. a warehouse robot that finds new routes around obstacles.
1 mark: makes its own decisions/reasons. 1 mark: learns/adapts, with an example.
Q28Automated systems · Core · Paper 1 style · 4 marks
(a)An automated system is used to control the lights on a motorway. State the three main components of an automated system.[3]
(b)Give one advantage of using an automated system to control the motorway lights.[1]
(b) Any one, e.g. lights switch on/off at the right time automatically, saving energy; no human needed to monitor; works 24 hours; faster response to changing light levels.
1 mark.
Q29Automated systems · Standard · Paper 1 style · 4 marks
(a)A greenhouse in Murree uses an automated system to control watering. Describe how the system keeps the soil moist.[4]
Show answer
(a) A moisture sensor measures the moisture level of the soil continuously; the reading is converted to digital (ADC) and sent to the microprocessor; the microprocessor compares the reading with a stored pre-set value; if the soil is too dry, it sends a signal to an actuator to open the valve/turn on the pump; when the reading reaches the pre-set value the valve is closed; the process repeats.
1 mark each for any four points.
Q30Automated systems · Challenge · Paper 1 style · 4 marks
(a)A weather-monitoring station on a remote mountain uses an automated system that records readings and sends them to a forecasting centre. Explain two advantages and two disadvantages of using an automated system rather than human observers.[4]
Show answer
(a)Advantages (any two): it can take readings continuously, day and night, without breaks; it works in dangerous or remote places where it would be unsafe/expensive for people to stay; readings are more accurate and consistent with no human error; data is sent immediately to the centre.Disadvantages (any two): expensive to set up and maintain; if a sensor fails it may send wrong data until someone visits; it needs a reliable power supply and communication link; it cannot react to unusual events it was not programmed for; risk of hacking/cyber attack.
1 mark for each advantage (max 2) and each disadvantage (max 2), applied to the weather station.
Similar questions
Questions that test the same subtopic in different years, grouped so you can work on one skill at a time.
References only: find each question in its paper (official link where available, otherwise through your school).
Artificial intelligence, expert systems and machine learning7 questions · 7/9 sessions
