Paper 1: Written examination

In 1997, IBM's Deep Blue became the first computer to beat a reigning world champion in a match at which game?

Which of these is an example of rule-based (symbolic) AI rather than machine learning?

What is an algorithm?

Which statement best describes a function in programming?

How does a tool such as Teachable Machine train an image model?

In machine learning, what is a feature?

What is overfitting?

What does the inference engine in a rule-based system do?

An AI agent is placed in a maze. What is its goal state?

Which of these is an environmental concern about AI?

An AI model is trained on 50 000 chest X-rays to detect pneumonia. The X-rays are labelled. Explain what "labelled" means in this context.

Describe how the developers should test the model in 11(a).

Most of the X-rays came from one hospital in one country. Explain a possible problem with this.

Explain why a doctor should still check the AI's diagnosis.

An agent must find a route through this grid from S to E. It can move Up, Down, Left or Right one square at a time. # is a wall.

     col1 col2 col3 col4row1  S    .    .    #row2  #    #    .    #row3  .    .    .    .row4  #    E    #    .

Write the shortest sequence of moves from S to E.

State the number of moves in the shortest route in 12(a).

A different maze agent uses this routine:

WHILE not at exit    IF path ahead is clear THEN        move forward    ELSE        turn right    ENDIFENDWHILE

Explain one problem with this routine.

Suggest how the routine in 12(c) could be improved.

A bank uses a rule-based system to make loan decisions. R1: IF income ≥ £20 000 AND existing debt < £5 000 THEN approve R2: IF income ≥ £20 000 AND existing debt ≥ £5 000 THEN refer to a human R3: IF income < £20 000 THEN reject Complete the decision for each applicant.

Applicant	Income	Existing debt	Decision
A	£25 000	£2 000	?
B	£18 000	£0	?
C	£40 000	£9 000	?

Identify the condition and the action in rule R3.

The bank replaces the rules with a machine learning model trained on 20 years of its past loan decisions. Explain how this could lead to discrimination.

def classify_temp(t):    if t < 10:        return "cold"    elif t < 25:        return "mild"    else:        return "hot" readings = [8, 25, 17]for r in readings:    print(classify_temp(r))

State the output of this program.

Identify the parameter of the function `classify_temp`.

Explain one benefit of using a function in this program.

State the value returned by `classify_temp(10)` and explain why 10 is an important value to test.

Write a function `is_freezing(t)` that returns `True` if the temperature is 0 or below and `False` otherwise.

A student uses Teachable Machine to build a model that recognises whether an item held up to a webcam is recyclable or not recyclable. Describe the steps the student should follow to build and test the model.

The model often classifies crisp packets as recyclable foil. Suggest two ways to improve it.

Explain the difference between classification and clustering.

A warehouse uses autonomous robots to move shelves. Explain how the robots use sensors to avoid collisions.

Explain one benefit and one drawback for the business of using autonomous robots in its warehouse.

A worker steps into a robot's path. Describe how the robot should respond.

Generative AI tools can create images in the style of a particular artist. Many were trained on images found online. Explain two legal issues this raises.

Some police forces use live facial recognition cameras in public places to find wanted people. Discuss whether police should be allowed to use this technology.

A school is considering using an AI tutor to replace some lessons, so that each student learns at their own pace. Evaluate this proposal.