Paper 1: Written examination

In which year was the term 'artificial intelligence' first used, at a summer workshop at Dartmouth College?

Which of these describes an 'AI winter'?

What is printed by this code?

x = 7y = 3print(x % y)

Deciding whether an email is spam or not spam is an example of which type of machine learning task?

What does pattern recognition mean in machine learning?

A self-driving car's model was trained mostly on photos taken in daylight. It performs poorly at night. What is the best explanation?

In a decision tree, where are the final outcomes (decisions) shown?

Which statement best describes an autonomous system?

What is meant by a 'black box' AI system?

What is a deepfake?

A music streaming app recommends songs to its users. State two types of data the app could collect about a user to make recommendations.

Explain how the app could use machine learning to recommend new songs.

The app keeps recommending songs from the same genre. Explain one disadvantage of this for the user.

Describe one privacy concern about the data the app collects.

This program checks a password.

1  password = input("Enter password: ")2  attempts = 13  while password != "Plasmo7" and attempts < 3:4      password = input("Try again: ")5      attempts = attempts + 16  if password == "Plasmo7":7      print("Access granted")8  else:9      print("Locked out")

State the line number where iteration begins.

State the data type of the variable `attempts` in 12(a).

A user enters `abc`, then `123`, then `xyz`. State the output of the program in 12(a) and the final value of `attempts`.

Explain why the keyword `and` is used on line 3.

Using line numbers from the program, explain the difference between sequence and selection.

Modify the program so that, after each wrong password, it outputs how many attempts are left. Write your answer in Python or pseudocode.

A school helpdesk chatbot is rule-based. One of its rules is: IF message contains "timetable" THEN reply with link to timetable Write two more IF–THEN rules this chatbot could use.

Explain two limitations of a rule-based chatbot compared with a chatbot that uses machine learning.

The school decides to train a machine learning chatbot on past helpdesk messages. Describe how the data should be prepared before training.

A smart traffic light at a junction has a road sensor that counts the cars waiting on the side road and a timer that measures how long the first car has waited. The main road normally has a green light. The side road should get a green light if 5 or more cars are waiting or a car has waited for more than 60 seconds. Write an algorithm, in pseudocode or Python, for this routine.

Explain how using sensor input makes this traffic light more efficient than one on a fixed timer.

Describe how the routine could be tested in a simulated environment before it is used on a real road.

A model predicts whether a piece of fruit is ripe. It is tested on new fruit:

	Predicted ripe	Predicted not ripe
Actually ripe	42	8
Actually not ripe	6	44

State how many pieces of fruit were tested.

Calculate the accuracy of the model in 15(a). Show your working.

A supermarket wants to avoid selling unripe fruit. Explain which type of error in the table matters most to the supermarket.

Explain why the images used to train this model should include fruit photographed in different lighting and from different angles.

An AI system stores knowledge as a semantic network: - Penguin —is a→ Bird - Bird —has→ Feathers - Bird —can→ Fly - Penguin —cannot→ Fly - Penguin —lives in→ Antarctica Explain what a semantic network is.

Using the network in 16(a), explain how the AI can conclude that penguins have feathers, even though this is not stored directly.

Explain why the network needs the link "Penguin cannot Fly".

A self-driving car is involved in a crash. Explain why it can be difficult to decide who is responsible.

Explain why transparency is important for AI systems such as self-driving cars.

Many students now have access to generative AI tools that can write text and answer questions. Discuss the benefits and risks of students using generative AI for their schoolwork.

"Artificial intelligence will make the world a better place." Evaluate this statement. You should consider both optimistic and cautious views about the future of AI.