A Level H1 General Paper Practice Paper 3

TuitionGoWhere Practice Paper — General Paper H1 A-Level

Practice Paper 2 — Comprehension: Answer Key (Version 3 of 5)

Section A: Short-Answer Questions (25 marks)

Question 1 (2 marks)

Answer: Two advantages of using AI in diagnostic radiology are:

1. AI can detect anomalies (such as tumours, fractures, and lesions) with a level of precision that rivals or surpasses that of experienced radiologists. (1 mark)
2. AI augments clinicians' capabilities, allowing them to focus on complex cases while routine screenings are handled efficiently by machines. (1 mark)

Marking Notes:

• Award 1 mark for each valid advantage, up to 2 marks.
• Accept paraphrased versions of the above points.
• Do not award marks for direct lifting of phrases such as "improving accuracy" without context.

Question 2 (2 marks)

Answer: The phrase "garbage in, garbage out" means that AI systems are only as reliable as the data they are trained on. (1 mark) If the training data is flawed, biased, or unrepresentative, the outputs produced by the AI will also be flawed or unreliable. (1 mark)

Marking Notes:

• Award 1 mark for explaining the general principle (quality of input determines quality of output).
• Award 1 mark for applying it to the context of AI and training data.
• Students must use their own words. Direct lifting of "AI systems are only as reliable as the data on which they are trained" should not be awarded the second mark.

Question 3 (2 marks)

Answer: The word "tempered" suggests that the author's attitude towards AI diagnostics is cautious and measured. (1 mark) While the author acknowledges the promise of AI, they believe enthusiasm should be balanced with an awareness of its limitations. (1 mark)

Marking Notes:

• Award 1 mark for identifying the cautious/qualified nature of the attitude.
• Award 1 mark for explaining that the author does not dismiss AI but urges balance.
• Accept synonyms such as "moderated," "balanced," or "qualified."

Question 4 (3 marks)

Answer: Two concerns about the reliability of AI diagnostic systems are: (a) Bias in training data: If training datasets underrepresent certain ethnic groups or age brackets, the AI may produce biased outcomes that are less accurate for those populations. (1.5 marks) (b) Inaccuracy across demographics: Research has shown that some diagnostic algorithms perform significantly less accurately for patients of African and Asian descent, raising concerns about equity in AI-driven healthcare. (1.5 marks)

Marking Notes:

• Award 1.5 marks for each concern: 1 mark for identifying the concern, 0.5 marks for explaining it in the student's own words.
• Both concerns are closely related but distinct: the first addresses the cause (skewed data), the second addresses the consequence (differential accuracy). Award marks for both if the student distinguishes them clearly.
• If a student gives the same concern twice in different words, award marks only once.

Question 5 (2 marks)

Answer: The author provides two pieces of evidence:

1. Traditional drug development takes over a decade and costs upwards of $2.6 billion per drug, whereas AI can identify promising drug candidates in a fraction of the time. (1 mark)
2. During the COVID-19 pandemic, companies like Moderna used AI to optimise mRNA sequences in record time, accelerating vaccine development. (1 mark)

Marking Notes:

• Award 1 mark for each piece of evidence, up to 2 marks.
• Students may paraphrase the evidence.

Question 6 (2 marks)

Answer: In this context, "serendipitous" refers to the element of chance or unexpected discovery in scientific research. (1 mark) The author suggests that AI, which relies on algorithmic predictions, may not be able to replicate the accidental or unplanned breakthroughs that have historically driven scientific progress. (1 mark)

Marking Notes:

• Award 1 mark for the definition in context.
• Award 1 mark for explaining the implication (AI may narrow the scope of research by relying on predictions rather than allowing for chance discoveries).

Question 7 (2 marks)

Answer: The author means that the idea of informed consent becomes unclear and complicated when patient data is used to train AI systems. (1 mark) Patients may not fully understand or agree to how their data is being repurposed, making it difficult to say they have truly consented. (1 mark)

Marking Notes:

• Award 1 mark for explaining "murky" as unclear/complicated.
• Award 1 mark for applying it to the context of informed consent and data use.
• Students must paraphrase; direct lifting of "patients may not fully understand how their information is being used" should not receive full credit.

Question 8 (3 marks)

Answer: The three main ethical concerns are: (a) Data privacy: AI systems require enormous volumes of patient data, and patients may not fully understand how their information is used or shared. (1 mark) (b) Accountability: It is unclear who is responsible when an AI system causes harm — the developer, the hospital, or the clinician. (1 mark) (c) Erosion of the doctor-patient relationship: Over-reliance on AI could reduce patients to data points and weaken the empathy and trust that define good medical practice. (1 mark)

Marking Notes:

• Award 1 mark for each valid concern.
• Accept alternative phrasings that capture the same ideas.

Question 9 (3 marks)

Answer: The author includes this quotation to reinforce the idea that medicine is fundamentally a human endeavour that requires both scientific knowledge and empathy. (1.5 marks) By quoting Gawande, the author lends authority to the argument that AI, however sophisticated, cannot replace the human qualities — empathy, trust, and clinical judgement — that define great doctors. (1.5 marks)

Marking Notes:

• Award up to 1.5 marks for identifying the purpose of the quotation (to support the argument about the human dimension of medicine).
• Award up to 1.5 marks for explaining how the quotation functions (appeal to authority, reinforcing the limits of AI).
• A well-developed answer may also note that the author uses Gawande's credibility to acknowledge AI's utility while still emphasising its limitations.

Question 10 (4 marks)

Answer: The sentence uses a contrast structure ("getting it right… getting it wrong") to juxtapose the best-case and worst-case outcomes of AI in healthcare, emphasising the magnitude of the stakes. (1.5 marks) The phrase "The stakes could not be higher" is a hyperbole that underscores the urgency and importance of the issue. (1 mark) The language choices — "save millions of lives" versus "deepen existing inequalities" and "undermine trust in medicine" — create a stark contrast between a hopeful and a cautionary vision, compelling the reader to recognise that the consequences of failure are not merely technical but deeply human and societal. (1.5 marks)

Marking Notes:

• Award 1.5 marks for identifying and explaining the contrast structure.
• Award 1 mark for identifying the hyperbole and its effect.
• Award 1.5 marks for analysing specific language choices and their impact.
• Award a maximum of 3 marks if the student identifies structural and language features but does not explain their effect on the reader.

Section B: Vocabulary in Context (5 marks)

Question 11 (1 mark) Answer: B (enhance)

• "Augment" means to increase or enhance, not to replace or diminish.

Question 12 (1 mark) Answer: B (argue)

• "Contend" in this context means to assert or argue a point, not to admit or concede.

Question 13 (1 mark) Answer: C (wisely)

• "Judiciously" means with good judgement or wisdom, i.e., using something sensibly and thoughtfully.

Question 14 (1 mark) Answer: C (significant and concerning)

• "Loom large" means something is prominent, imposing, or a source of concern.

Question 15 (1 mark) Answer: B (gradually weaken)

• "Erode" means to gradually wear away or diminish, not to destroy completely or strengthen.

Section C: Summary (10 marks)

Question 16

Model Summary (for reference — within 120 words):

AI in healthcare faces several limitations and concerns. Diagnostic AI systems are limited by the quality of their training data; if datasets are unrepresentative, biased outcomes may result, particularly for underrepresented ethnic groups. In drug development, while AI accelerates the process, critics warn it cannot replicate the chance discoveries that drive scientific breakthroughs. Ethically, the use of vast patient data raises privacy concerns, as patients may not fully consent to how their information is used. Accountability is also unclear when AI errors cause harm. Furthermore, over-reliance on AI risks eroding the empathy and human connection central to the doctor-patient relationship, qualities that no algorithm can replicate.

(Approximately 110 words)

Marking Scheme:

Content Points (award 1 mark each, up to 7 marks):

1. AI systems are only as reliable as their training data.
2. Skewed/unrepresentative training data leads to biased outcomes.
3. Algorithms may be less accurate for certain ethnic groups (e.g., African and Asian descent).
4. AI cannot replicate serendipitous/chance scientific discoveries.
5. Data privacy concerns arise from the use of large volumes of patient data.
6. Informed consent is unclear when data is repurposed for AI.
7. Accountability is unresolved when AI errors cause harm.
8. Over-reliance on AI may erode the doctor-patient relationship/empathy.

Language Marks (award as follows):

• 3 marks: Excellent paraphrasing, fluent continuous prose, well within word limit.
• 2 marks: Good paraphrasing with minor lifting, generally fluent.
• 1 mark: Significant lifting from the passage, limited paraphrasing.
• 0 marks: Wholly lifted from the passage or exceeds word limit significantly.

Common Mistakes:

• Including information from paragraph 1 (introduction) or paragraph 6 (conclusion) — these are outside the specified paragraphs.
• Exceeding the 120-word limit.
• Failing to use own words (direct lifting).

Section D: Application Question (10 marks)

Question 17 (3 marks)

Answer: Two potential benefits of Singapore's AI healthcare initiative are: (a) Reduced waiting times: The passage notes that AI can handle routine screenings efficiently, and the scenario states the initiative aims to reduce diagnosis waiting times by 40%. (1.5 marks) (b) Addressing healthcare workforce shortages: The passage explains that AI augments clinicians' capabilities, and the scenario highlights that the initiative will help address the shortage of healthcare professionals as Singapore's population ages. (1.5 marks)

Marking Notes:

• Award 1.5 marks for each benefit: 1 mark for identifying the benefit, 0.5 marks for linking it to ideas from the passage.
• Accept other valid benefits drawn from the passage (e.g., improved diagnostic accuracy, cost savings).

Question 18 (3 marks)

Answer: Singapore's initiative could affect the doctor-patient relationship in two ways. Positively, the passage notes that AI can free physicians from administrative burdens, allowing them to devote more time to meaningful patient interaction. (1 mark) The scenario echoes this, with the Health Minister assuring that AI will "empower" doctors to deliver faster, more accurately care. (0.5 marks) However, the passage warns that over-reliance on AI could erode empathy and reduce patients to data points, (1 mark) a concern echoed by the Singapore Medical Association in the scenario, which fears doctors may become over-reliant on AI and lose independent clinical judgement. (0.5 marks)

Marking Notes:

• Award up to 1.5 marks for each side of the argument (positive and negative effects).
• Award 0.5 marks for each reference to the passage or scenario.
• Answers must draw on both the passage and the scenario for full marks.

Question 19 (2 marks)

Answer: Two measures Singapore should implement are: (a) Stronger data protection legislation to address public concerns about the privacy and security of medical data in the national digital database. (1 mark) (b) Clear legal frameworks on liability to determine who is responsible when AI-assisted diagnoses lead to errors, as called for by MPs in the scenario and discussed in the passage. (1 mark)

Marking Notes:

• Award 1 mark for each valid measure linked to concerns raised in the passage or scenario.
• Accept other reasonable measures (e.g., public education on how AI works, ensuring diverse training data, mandatory human oversight of AI diagnoses).

Question 20 (2 marks)

Answer: (Accept any well-reasoned position. Sample answer below.)

Sample Position (Benefits outweigh risks): The benefits of AI in healthcare outweigh the risks because, as the passage notes, AI has the potential to save millions of lives through faster, more accurate diagnosis and accelerated drug development. (1 mark) The scenario shows that Singapore's initiative could reduce waiting times by 40% and address critical workforce shortages, which are pressing needs in an ageing society. Provided that robust regulations and human oversight are in place, the risks can be effectively managed. (1 mark)

Sample Position (Risks outweigh benefits): The risks of AI in healthcare may outweigh the benefits because, as the passage highlights, biased algorithms could deepen existing inequalities, and the erosion of the doctor-patient relationship could fundamentally undermine the quality of care. (1 mark) The scenario reveals that only 34% of Singaporeans understand how AI works, and nearly half are concerned about data privacy, suggesting that public trust has not yet been earned. Without addressing these concerns, the initiative risks doing more harm than good. (1 mark)

Marking Notes:

• Award 1 mark for a clear position.
• Award 1 mark for a reason supported by evidence from the passage or scenario.
• Accept either position (for or against) as long as it is well-supported.

END OF ANSWER KEY

Marks Summary: