IELTS Reading: AI Trong Bảo Tồn Môi Trường – Đề Thi Mẫu Có Đáp Án Chi Tiết

Mở Bài

Trí tuệ nhân tạo (AI) đang cách mạng hóa nhiều lĩnh vực, và bảo tồn môi trường không phải là ngoại lệ. Chủ đề “How Is AI Being Used In Environmental Conservation?” ngày càng xuất hiện nhiều trong các đề thi IELTS Reading, phản ánh xu hướng quan tâm toàn cầu về công nghệ xanh và phát triển bền vững. Với hơn 20 năm kinh nghiệm giảng dạy IELTS, tôi nhận thấy chủ đề này thường xuất hiện với tần suất cao, đặc biệt trong các bài thi từ năm 2020 trở đi.

Bài viết này cung cấp một bộ đề thi IELTS Reading hoàn chỉnh với 3 passages theo đúng format thi thật, từ độ khó Easy đến Hard. Bạn sẽ được luyện tập với 40 câu hỏi đa dạng, bao gồm Multiple Choice, True/False/Not Given, Matching Headings, Summary Completion và nhiều dạng khác. Đặc biệt, bài viết kèm theo đáp án chi tiết với giải thích cụ thể, từ vựng quan trọng và các tips làm bài thực chiến.

Đề thi này phù hợp với học viên từ band 5.0 trở lên, giúp bạn làm quen với chủ đề công nghệ-môi trường và nâng cao kỹ năng đọc hiểu học thuật một cách hiệu quả.

1. Hướng Dẫn Làm Bài IELTS Reading

Tổng Quan Về IELTS Reading Test

IELTS Reading Test bao gồm 3 passages với độ dài và độ khó tăng dần. Bạn có 60 phút để hoàn thành 40 câu hỏi. Không có thời gian thêm để chuyển đáp án sang phiếu trả lời, vì vậy quản lý thời gian là yếu tố then chốt.

Phân bổ thời gian khuyến nghị:

• Passage 1: 15-17 phút (13 câu hỏi – độ khó Easy)
• Passage 2: 18-20 phút (13 câu hỏi – độ khó Medium)
• Passage 3: 23-25 phút (14 câu hỏi – độ khó Hard)

Lưu ý dành thời gian kiểm tra lại đáp án, đặc biệt chú ý lỗi chính tả khi viết câu trả lời dạng điền từ.

Các Dạng Câu Hỏi Trong Đề Này

Đề thi mẫu này bao gồm đầy đủ các dạng câu hỏi phổ biến nhất trong IELTS Reading:

1. Multiple Choice – Lựa chọn đáp án đúng từ các phương án cho sẵn
2. True/False/Not Given – Xác định tính đúng sai của thông tin so với passage
3. Matching Headings – Ghép tiêu đề phù hợp với các đoạn văn
4. Summary Completion – Hoàn thành đoạn tóm tắt bằng từ trong bài
5. Matching Features – Ghép thông tin với các đối tượng được đề cập
6. Short-answer Questions – Trả lời câu hỏi ngắn dựa trên thông tin trong bài

Mỗi dạng câu hỏi yêu cầu kỹ năng đọc và chiến lược làm bài khác nhau, được thiết kế để đánh giá toàn diện năng lực Reading của thí sinh.

Học viên đang luyện tập IELTS Reading với chủ đề AI trong bảo tồn môi trường trên máy tính

2. IELTS Reading Practice Test

PASSAGE 1 – Artificial Intelligence: A New Tool for Wildlife Protection

Độ khó: Easy (Band 5.0-6.5)

Thời gian đề xuất: 15-17 phút

A. The use of artificial intelligence (AI) in wildlife conservation has emerged as one of the most promising developments in recent years. Traditional methods of monitoring endangered species often require substantial human resources and can be both time-consuming and expensive. However, AI technologies are now enabling conservationists to track, identify, and protect wildlife more efficiently than ever before. From recognizing individual animals in photographs to predicting poaching activities, AI is transforming how we approach environmental protection.

B. One of the most successful applications of AI in conservation involves camera trap technology. Thousands of cameras are positioned in forests and natural habitats worldwide, capturing millions of images of wildlife. Previously, researchers had to manually review each photograph, a process that could take months or even years. Today, machine learning algorithms can automatically identify species, count individuals, and even recognize specific animals by their unique markings. For example, a system called Wildlife Insights, developed by Conservation International, processes camera trap images from over 350 different projects across the globe. This platform uses AI to identify animals in photos with more than 96% accuracy, completing in minutes what would take humans weeks.

C. Another innovative use of AI is in combating illegal wildlife trade. Organizations like TRAFFIC and the International Fund for Animal Welfare have implemented AI systems that scan online marketplaces for advertisements selling protected species. These algorithms can detect suspicious listings by analyzing text, images, and seller patterns. The technology has proven particularly effective in identifying the sale of ivory, rhino horn, and exotic pets on social media platforms and e-commerce websites. In one case, an AI system monitored over 30,000 online advertisements per month and successfully identified more than 500 illegal wildlife products that were subsequently removed from sale.

D. Acoustic monitoring represents yet another frontier where AI is making significant contributions. Many animals, from whales to tropical birds, produce distinctive sounds that can be used to track their populations and behaviors. AI-powered acoustic sensors can be deployed in remote locations to continuously record environmental sounds. Advanced pattern recognition software then analyzes these recordings to identify specific species calls, even filtering out background noise like wind or rain. Researchers studying the Brazilian rainforest, for instance, used acoustic AI to discover that certain frog species were far more widespread than previously believed, leading to revised conservation strategies.

E. The prediction of environmental threats is another area where AI demonstrates remarkable potential. By analyzing historical data on factors such as weather patterns, human activity, and animal movements, predictive models can forecast where poaching incidents or habitat destruction are most likely to occur. PAWS (Protection Assistant for Wildlife Security), developed by researchers at Harvard University, uses game theory and machine learning to help rangers optimize their patrol routes in protected areas. Parks using PAWS have reported up to a 40% reduction in poaching activities, as the system enables more strategic deployment of limited security resources.

F. Despite these successes, the integration of AI into conservation work faces several challenges. The technology requires reliable internet connectivity and electrical power, which may not be available in remote conservation areas. Additionally, developing and training AI systems demands significant technical expertise and financial investment. There are also concerns about data privacy and the potential misuse of tracking technologies. Some critics argue that over-reliance on technology might reduce direct human engagement with nature, which is essential for building public awareness and support for conservation causes.

G. Nevertheless, the future of AI in environmental conservation appears bright. As technology becomes more affordable and accessible, smaller conservation organizations are beginning to adopt these tools. Partnerships between tech companies and environmental groups are expanding, bringing together expertise from both fields. Recent developments in edge computing – where data processing occurs directly on devices rather than in distant data centers – promise to make AI conservation tools more practical for remote locations. With continued innovation and collaboration, AI is poised to become an indispensable ally in the global effort to protect our planet’s biodiversity.

---

Questions 1-13

Questions 1-5: Multiple Choice

Choose the correct letter, A, B, C, or D.

1. What is the main advantage of using AI in wildlife conservation mentioned in paragraph A?
A. It completely replaces human conservationists
B. It makes conservation work more efficient
C. It is less expensive than traditional methods in all cases
D. It can predict animal behavior perfectly

2. According to paragraph B, the Wildlife Insights system:
A. Takes weeks to process camera trap images
B. Was developed by Harvard University
C. Achieves over 96% accuracy in species identification
D. Only works in certain types of forests

3. The AI system monitoring online wildlife trade mentioned in paragraph C:
A. Checks approximately 30,000 advertisements monthly
B. Has completely eliminated illegal wildlife sales
C. Only works on social media platforms
D. Identifies sellers by their personal information

4. What is acoustic monitoring used for, according to paragraph D?
A. Recording only whale sounds
B. Tracking animal populations through their sounds
C. Eliminating background noise in forests
D. Discovering new species exclusively

5. PAWS (Protection Assistant for Wildlife Security) has helped reduce poaching by:
A. Increasing the number of park rangers
B. Installing more security cameras
C. Optimizing patrol routes using AI
D. Capturing all poachers in protected areas

Questions 6-9: True/False/Not Given

Do the following statements agree with the information given in the passage?

Write:

• TRUE if the statement agrees with the information
• FALSE if the statement contradicts the information
• NOT GIVEN if there is no information on this

6. Traditional wildlife monitoring methods require significant human effort and resources.

7. Camera trap technology was first invented by Conservation International.

8. AI systems can identify individual animals by their unique markings.

9. All conservation organizations worldwide now use AI technology.

Questions 10-13: Summary Completion

Complete the summary below using NO MORE THAN TWO WORDS from the passage for each answer.

AI technology faces several challenges in conservation work. Remote areas often lack reliable (10) ____ and electrical power. Developing AI systems requires considerable (11) ____ and financial resources. Some people worry about (12) ____ issues and potential misuse of tracking systems. Critics also suggest that too much reliance on technology might decrease direct (13) ____ with the natural world.

---

PASSAGE 2 – Machine Learning Algorithms Transform Ecosystem Management

Độ khó: Medium (Band 6.0-7.5)

Thời gian đề xuất: 18-20 phút

A. The convergence of artificial intelligence and environmental science has catalyzed a paradigm shift in how ecosystems are monitored, analyzed, and managed. Unlike conventional conservation approaches that rely heavily on manual observation and periodic surveys, AI-driven systems provide continuous, real-time data collection and analysis at scales previously deemed impossible. This technological evolution is particularly crucial given the accelerating rate of biodiversity loss and climate change impacts that demand rapid, evidence-based responses from conservation practitioners worldwide.

B. Satellite imagery analysis powered by deep learning algorithms exemplifies this transformation. Organizations such as Global Forest Watch utilize convolutional neural networks (CNNs) to process terabytes of satellite data daily, detecting deforestation activities with unprecedented precision. These algorithms can distinguish between natural forest loss from events like wildfires and anthropogenic destruction caused by illegal logging or agricultural expansion. The system operates on a near-real-time basis, issuing alerts within days of detecting forest clearance, enabling authorities to respond before significant damage occurs. In the Brazilian Amazon, this technology has contributed to a 15% increase in successful interventions against illegal deforestation operations.

C. The application of AI extends beyond terrestrial ecosystems into marine conservation. Autonomous underwater vehicles (AUVs) equipped with AI-powered imaging systems now survey coral reef health across vast oceanic areas. These systems employ computer vision techniques to assess coral bleaching severity, identify invasive species, and monitor fish populations. What traditionally required hundreds of diving hours by marine biologists can now be accomplished in a fraction of the time with greater consistency. The Great Barrier Reef Marine Park Authority has deployed such technology to create comprehensive baseline assessments of reef conditions, enabling more targeted restoration efforts.

D. Predictive modeling represents another frontier where machine learning demonstrates exceptional utility. By integrating multidimensional datasets – including climate variables, land use patterns, population demographics, and historical conservation outcomes – AI algorithms can forecast future environmental conditions and species distribution shifts. Researchers at Stanford University developed a model that predicts habitat suitability for endangered species under various climate change scenarios with 87% accuracy. These predictions inform strategic land acquisition decisions by conservation organizations, ensuring protection of areas that will remain viable habitats even as climate patterns evolve.

E. The realm of genomic conservation has similarly been revolutionized by artificial intelligence. Conservation geneticists now employ machine learning algorithms to analyze genetic diversity within endangered populations, identifying individuals that should be prioritized for breeding programs to maximize genetic health. AI systems can process whole-genome sequences exponentially faster than traditional methods, detecting subtle genetic markers associated with disease resistance or adaptive traits. This capability proved instrumental in conservation efforts for the California condor, where AI-guided genetic management has significantly improved population viability.

F. Natural language processing (NLP), a branch of AI focused on understanding human language, contributes to conservation in unexpected ways. Researchers have developed systems that automatically scan scientific literature, government reports, and news articles to track emerging environmental threats and identify conservation opportunities. These text-mining algorithms can detect patterns and connections that might escape human notice, such as correlations between infrastructure development projects and increased threats to specific species. Environmental organizations use these insights to prioritize advocacy efforts and allocate resources more effectively.

G. However, the implementation of AI in conservation is not without methodological challenges and ethical considerations. Algorithmic bias poses a significant concern – if training data predominantly comes from well-studied ecosystems in developed nations, the resulting models may perform poorly in under-researched regions where conservation needs are often most acute. Furthermore, the computational resources required for sophisticated AI systems raise questions about carbon footprints and energy consumption. There is an inherent irony in using energy-intensive technologies to address environmental problems. Experts advocate for the development of more energy-efficient algorithms and powering AI infrastructure with renewable energy sources.

H. The issue of technological accessibility also merits attention. While large international NGOs can afford cutting-edge AI systems, small grassroots organizations in biodiversity hotspots often lack the financial and technical capacity to implement these tools. This digital divide risks creating a two-tiered conservation landscape where well-funded projects benefit from advanced technology while resource-poor initiatives continue with outdated methods. Addressing this disparity requires intentional efforts to develop open-source platforms, provide training programs, and establish technology transfer mechanisms that democratize access to AI conservation tools.

I. Looking forward, the integration of AI with other emerging technologies promises even greater conservation impact. The combination of AI with Internet of Things (IoT) sensors, blockchain for supply chain verification, and drone technology creates comprehensive environmental monitoring ecosystems. For instance, smart sensor networks in protected areas can detect unusual animal movements, environmental changes, or human intrusions, with AI systems automatically analyzing data and alerting rangers only when intervention is necessary. Such integrated approaches maximize efficiency while minimizing the need for continuous human monitoring, allowing conservation personnel to focus on strategic decision-making and community engagement.

---

Questions 14-26

Questions 14-18: Yes/No/Not Given

Do the following statements agree with the claims of the writer in the passage?

Write:

• YES if the statement agrees with the claims of the writer
• NO if the statement contradicts the claims of the writer
• NOT GIVEN if it is impossible to say what the writer thinks about this

14. AI-driven conservation systems provide data collection capabilities that were impossible with traditional methods.

15. All deforestation detected by satellite systems is caused by human activities.

16. Autonomous underwater vehicles can survey coral reefs more consistently than human divers.

17. The Stanford University model for predicting habitat suitability is 100% accurate.

18. Machine learning algorithms for analyzing genetic diversity work more slowly than traditional methods.

Questions 19-22: Matching Headings

The passage has nine paragraphs, A-I. Choose the correct heading for paragraphs E, F, G, and H from the list of headings below.

List of Headings:
i. The role of AI in genetic conservation efforts
ii. Commercial applications of conservation technology
iii. Challenges related to algorithmic bias and energy use
iv. The future combination of multiple technologies
v. Using AI to analyze written information for conservation
vi. Unequal access to conservation technology
vii. Traditional methods of species monitoring
viii. Criticism of AI in environmental work

19. Paragraph E
20. Paragraph F
21. Paragraph G
22. Paragraph H

Questions 23-26: Summary Completion

Complete the summary below using NO MORE THAN THREE WORDS from the passage for each answer.

Global Forest Watch uses (23) ____ to analyze large amounts of satellite data every day. These algorithms can tell the difference between natural forest loss and (24) ____ caused by humans. In Brazil, this has led to a 15% improvement in stopping (25) ____. The system can send alerts within days, allowing authorities to respond quickly. Similarly, underwater vehicles equipped with AI assess (26) ____ and monitor marine life across large ocean areas.

---

Hệ thống AI phân tích hình ảnh vệ tinh giám sát rừng nhiệt đới để phát hiện phá rừng trái phép

PASSAGE 3 – The Algorithmic Turn in Conservation Biology: Opportunities, Limitations, and Epistemological Implications

Độ khó: Hard (Band 7.0-9.0)

Thời gian đề xuất: 23-25 phút

A. The integration of artificial intelligence into conservation biology represents not merely a technological augmentation of existing practices but rather a fundamental epistemological transformation in how environmental knowledge is generated, validated, and operationalized. This algorithmic turn – characterized by the deployment of machine learning, neural networks, and computational modeling across diverse conservation contexts – necessitates critical examination of both its transformative potential and its inherent limitations. As conservation science increasingly relies on data-driven methodologies, questions arise regarding the relationship between algorithmic inference and ecological understanding, the role of human expertise in automated decision-making systems, and the broader implications for conservation governance and policy formulation.

B. Contemporary deep learning architectures have demonstrated remarkable capabilities in pattern recognition tasks that form the foundation of much conservation work. Residual neural networks (ResNets) and transformer models, originally developed for computer vision and natural language processing, have been repurposed for ecological applications with striking efficacy. For instance, research published in the Proceedings of the National Academy of Sciences documented a system utilizing attention mechanisms – a sophisticated neural network component – to identify individual endangered primates from camera trap imagery with accuracy surpassing that of experienced field researchers. The system achieved 98.7% precision across 15 different primate species, even under challenging conditions of occlusion, variable lighting, and partial visibility. Critically, the model’s feature extraction process revealed that it focused on morphological characteristics (such as facial structure and pelage patterns) that aligned with those used by human experts, suggesting a degree of biological validity in the algorithmic approach.

C. The application of reinforcement learning – an AI paradigm where algorithms learn optimal strategies through trial-and-error interaction with their environment – presents particularly intriguing possibilities for adaptive management scenarios. Unlike supervised learning approaches that require extensive labeled datasets, reinforcement learning agents can operate in environments with sparse feedback, gradually refining their strategies based on reward signals that encode conservation objectives. Researchers at the University of British Columbia developed a reinforcement learning system to optimize freshwater allocation in river systems that must balance competing demands: maintaining ecological flows for fish populations, supporting agricultural irrigation, and ensuring municipal water supply. The AI agent learned strategies that improved fish habitat conditions by 23% compared to existing water management policies, while maintaining agricultural and municipal supply within acceptable parameters. This success derived from the algorithm’s capacity to identify non-intuitive temporal patterns in water release timing that human managers had not recognized despite decades of experience.

D. However, the opacity of complex machine learning models poses significant epistemological and practical challenges for conservation science. Many high-performing deep neural networks function as “black boxes” – generating accurate predictions while providing limited insight into the causal mechanisms underlying those predictions. This interpretability deficit proves particularly problematic in conservation contexts where understanding why a species is declining or how an intervention works is often as important as predicting outcomes. The inability to extract mechanistic understanding from algorithmic models may impede the development of ecological theory and limit the transferability of AI-derived insights across different geographic or taxonomic contexts. Conservation biologists have begun addressing this challenge by employing explainable AI (XAI) techniques – methods designed to render algorithmic decision-making processes more transparent and interpretable. SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) represent two such approaches, providing post-hoc interpretations of complex model predictions by quantifying the contribution of individual input features to specific outcomes.

E. The data requirements of contemporary machine learning systems present additional constraints on their conservation applications. While AI algorithms excel at extracting patterns from large datasets, many conservation scenarios involve data-scarce environments where comprehensive observations are logistically or financially prohibitive. Endangered species, by definition, exist in limited numbers, providing few training examples for supervised learning algorithms. Transfer learning – leveraging models trained on abundant data from one context and fine-tuning them for related tasks with limited data – offers a partial solution. Conservation biologists have successfully applied models initially trained on millions of common species images to endangered species identification tasks with only hundreds of training examples. Nevertheless, concerns persist about distributional shift – the possibility that patterns learned from abundant species may not generalize to rare species with distinct ecological characteristics.

F. The socio-political dimensions of algorithmic conservation warrant equally rigorous scrutiny. The increasing reliance on AI-driven environmental monitoring and decision-making raises questions about technocratic governance and the marginalization of local ecological knowledge and indigenous wisdom. When sophisticated algorithms determine conservation priorities or predict environmental futures, there is risk of depoliticizing fundamentally political decisions about resource allocation and land use. Critics argue that framing conservation challenges as technical problems solvable through better algorithms obscures the underlying structural inequalities and power dynamics that drive environmental degradation. Furthermore, the concentration of AI expertise within elite institutions in wealthy nations risks perpetuating colonial patterns in conservation, where interventions are designed externally and imposed on biodiverse regions in the Global South with limited input from local communities.

G. Algorithmic accountability emerges as another critical concern. When AI systems inform consequential conservation decisions – such as anti-poaching operations that may result in armed confrontations, or land-use restrictions that affect local livelihoods – questions arise regarding responsibility for errors or unintended consequences. Legal and ethical frameworks for algorithmic accountability remain nascent, particularly in international conservation contexts spanning multiple jurisdictions. Who bears responsibility when an AI-guided enforcement operation goes wrong? How should trade-offs between conservation effectiveness and community impacts be adjudicated when algorithmic recommendations conflict with local interests? These questions lack clear answers, yet become increasingly urgent as AI adoption expands.

H. The energy consumption and carbon footprint of artificial intelligence present an paradoxical challenge for its application in environmental conservation. Training large-scale neural networks requires substantial computational resources; a 2019 study by researchers at the University of Massachusetts Amherst found that training a single large natural language processing model can emit approximately 284,000 kilograms of carbon dioxide – equivalent to the lifetime emissions of five automobiles. While conservation-specific AI applications typically employ smaller models with correspondingly lower environmental costs, the aggregate impact of expanding AI infrastructure merits consideration. This paradox highlights the importance of developing energy-efficient algorithms, utilizing renewable energy sources for computational infrastructure, and carefully evaluating whether the conservation benefits of AI applications justify their environmental costs.

I. Despite these challenges, the trajectory of AI in conservation appears oriented toward increasingly sophisticated and integrated applications. Emerging paradigms such as multi-agent systems – where multiple AI agents with different specializations collaborate on complex problems – promise to address some current limitations. Federated learning approaches enable training AI models across distributed datasets without requiring data centralization, potentially addressing both privacy concerns and the challenges of integrating diverse information sources. The development of hybrid systems that combine algorithmic processing with human expertise, rather than seeking to replace human judgment entirely, offers promising pathways for leveraging the complementary strengths of human and artificial intelligence. As these technologies mature, the critical task for conservation science lies not in uncritical adoption or wholesale rejection, but in thoughtfully integrating AI capabilities within ethical frameworks that prioritize both ecological integrity and social justice.

---

Questions 27-40

Questions 27-31: Multiple Choice

Choose the correct letter, A, B, C, or D.

27. According to paragraph A, the integration of AI in conservation biology represents:
A. A simple improvement in existing conservation techniques
B. A fundamental change in how conservation knowledge is created
C. The replacement of human conservationists with machines
D. A temporary trend in environmental science

28. The research on primate identification mentioned in paragraph B showed that:
A. AI systems cannot match human expert accuracy
B. The AI focused on different features than human experts
C. The system achieved 98.7% precision across 15 species
D. Camera trap imagery is unsuitable for AI analysis

29. The reinforcement learning system for freshwater allocation (paragraph C):
A. Required extensive labeled datasets to function
B. Only improved fish habitat conditions by 5%
C. Discovered water release patterns humans had not identified
D. Could not balance competing water demands

30. The “black box” problem in AI conservation (paragraph D) refers to:
A. The high cost of AI systems
B. The inability to understand how AI makes predictions
C. The color of computer equipment
D. The physical size of computing systems

31. What does paragraph H identify as paradoxical about AI in conservation?
A. AI is too expensive for most conservation projects
B. AI requires human operators who lack training
C. Using energy-intensive AI to solve environmental problems
D. AI cannot function in remote conservation areas

Questions 32-36: Matching Features

Match each limitation or challenge (Questions 32-36) with the correct aspect of AI conservation (A-H). You may use any letter more than once.

Aspects of AI Conservation:
A. Data requirements
B. Interpretability of results
C. Energy consumption
D. Socio-political dimensions
E. Algorithmic accountability
F. Transfer learning
G. Attention mechanisms
H. Reinforcement learning

32. The difficulty of obtaining enough training examples for rare species

33. Concerns about excluding local communities from decision-making

34. Unclear responsibility when AI-guided operations have negative outcomes

35. Models that generate accurate predictions without explaining why

36. The carbon emissions from training large neural networks

Questions 37-40: Short-answer Questions

Answer the questions below using NO MORE THAN THREE WORDS AND/OR A NUMBER from the passage for each answer.

37. What two XAI techniques are mentioned that help make AI decisions more transparent?

38. What term describes using models trained on one task for a related task with less data?

39. According to the passage, what might be obscured when conservation is framed as a technical problem?

40. What type of system combines multiple AI agents with different specializations?

---

3. Answer Keys – Đáp Án

PASSAGE 1: Questions 1-13

1. B
2. C
3. A
4. B
5. C
6. TRUE
7. NOT GIVEN
8. TRUE
9. FALSE
10. internet connectivity
11. technical expertise
12. data privacy
13. human engagement

PASSAGE 2: Questions 14-26

14. YES
15. NO
16. YES
17. NO
18. NO
19. i
20. v
21. iii
22. vi
23. convolutional neural networks / deep learning algorithms
24. anthropogenic destruction
25. illegal deforestation operations / illegal deforestation
26. coral bleaching severity / coral reef health

PASSAGE 3: Questions 27-40

27. B
28. C
29. C
30. B
31. C
32. A
33. D
34. E
35. B
36. C
37. SHAP and LIME
38. Transfer learning
39. structural inequalities / power dynamics
40. Multi-agent systems

---

4. Giải Thích Đáp Án Chi Tiết

Passage 1 – Giải Thích

Câu 1: B

• Dạng câu hỏi: Multiple Choice
• Từ khóa: main advantage, AI, wildlife conservation
• Vị trí trong bài: Đoạn A, dòng 2-4
• Giải thích: Bài đọc nói rõ “AI technologies are now enabling conservationists to track, identify, and protect wildlife more efficiently than ever before.” Từ “efficiently” được paraphrase thành “makes conservation work more efficient” trong đáp án B. Đáp án A sai vì AI không thay thế hoàn toàn con người. Đáp án C quá tuyệt đối (không phải “trong mọi trường hợp”). Đáp án D không được đề cập.

Câu 6: TRUE

• Dạng câu hỏi: True/False/Not Given
• Từ khóa: Traditional wildlife monitoring, significant human effort
• Vị trí trong bài: Đoạn A, dòng 1-2
• Giải thích: Bài viết nói “Traditional methods of monitoring endangered species often require substantial human resources and can be both time-consuming and expensive.” “Substantial human resources” = “significant human effort and resources”. Câu này đúng hoàn toàn với thông tin trong bài.

Câu 8: TRUE

• Dạng câu hỏi: True/False/Not Given
• Từ khóa: AI systems, identify individual animals, unique markings
• Vị trí trong bài: Đoạn B, dòng 4-5
• Giải thích: Bài đọc đề cập “machine learning algorithms can automatically identify species, count individuals, and even recognize specific animals by their unique markings.” Điều này khớp chính xác với câu hỏi.

Câu 10: internet connectivity

• Dạng câu hỏi: Summary Completion
• Từ khóa: Remote areas, lack
• Vị trí trong bài: Đoạn F, dòng 2
• Giải thích: Câu trong bài: “The technology requires reliable internet connectivity and electrical power, which may not be available in remote conservation areas.” Cần điền “internet connectivity” vào chỗ trống.

Câu 13: human engagement

• Dạng câu hỏi: Summary Completion
• Từ khóa: reduce direct, with natural world
• Vị trí trong bài: Đoạn F, dòng 5-6
• Giải thích: Bài viết nói “over-reliance on technology might reduce direct human engagement with nature”. Cụm “human engagement” chính xác cần điền vào chỗ trống.

Passage 2 – Giải Thích

Câu 14: YES

• Dạng câu hỏi: Yes/No/Not Given
• Từ khóa: AI-driven systems, data collection, impossible, traditional methods
• Vị trí trong bài: Đoạn A, dòng 2-4
• Giải thích: Tác giả khẳng định “AI-driven systems provide continuous, real-time data collection and analysis at scales previously deemed impossible.” Từ “previously deemed impossible” khớp với ý “were impossible with traditional methods”. Đây là quan điểm của tác giả nên chọn YES.

Câu 15: NO

• Dạng câu hỏi: Yes/No/Not Given
• Từ khóa: All deforestation, human activities
• Vị trí trong bài: Đoạn B, dòng 4-6
• Giải thích: Bài viết nói “These algorithms can distinguish between natural forest loss from events like wildfires and anthropogenic destruction caused by illegal logging”. Điều này chứng tỏ KHÔNG PHẢI tất cả phá rừng đều do con người gây ra, có cả nguyên nhân tự nhiên. Câu này mâu thuẫn với ý kiến tác giả nên chọn NO.

Câu 19: i

• Dạng câu hỏi: Matching Headings
• Đoạn văn: E
• Giải thích: Đoạn E tập trung vào “genomic conservation”, “breeding programs”, “genetic diversity” và “genetic management” – tất cả liên quan đến vai trò của AI trong bảo tồn di truyền. Tiêu đề “The role of AI in genetic conservation efforts” khớp hoàn hảo.

Câu 21: iii

• Dạng câu hỏi: Matching Headings
• Đoạn văn: G
• Giải thích: Đoạn G thảo luận về “algorithmic bias” và “energy consumption” cùng các “ethical considerations”. Tiêu đề “Challenges related to algorithmic bias and energy use” phản ánh chính xác nội dung đoạn này.

Câu 23: convolutional neural networks / deep learning algorithms

• Dạng câu hỏi: Summary Completion (NO MORE THAN THREE WORDS)
• Vị trí trong bài: Đoạn B, dòng 2-3
• Giải thích: “Organizations such as Global Forest Watch utilize convolutional neural networks (CNNs) to process terabytes of satellite data daily”. Cả hai cụm từ đều chấp nhận được vì đoạn văn đề cập “deep learning algorithms” và “convolutional neural networks (CNNs)” đều được sử dụng cho mục đích này.

Cảm biến acoustic AI ghi âm tiếng kêu động vật hoang dã trong rừng nhiệt đới

Passage 3 – Giải Thích

Câu 27: B

• Dạng câu hỏi: Multiple Choice
• Từ khóa: integration of AI, conservation biology, represents
• Vị trí trong bài: Đoạn A, dòng 1-2
• Giải thích: Câu đầu tiên của bài nói “represents not merely a technological augmentation of existing practices but rather a fundamental epistemological transformation in how environmental knowledge is generated”. Đáp án B “A fundamental change in how conservation knowledge is created” là paraphrase chính xác của câu này.

Câu 29: C

• Dạng câu hỏi: Multiple Choice
• Từ khóa: reinforcement learning system, freshwater allocation
• Vị trí trong bài: Đoạn C, dòng 8-10
• Giải thích: Bài viết nói “This success derived from the algorithm’s capacity to identify non-intuitive temporal patterns in water release timing that human managers had not recognized despite decades of experience.” Đáp án C “Discovered water release patterns humans had not identified” chính xác diễn đạt ý này.

Câu 30: B

• Dạng câu hỏi: Multiple Choice
• Từ khóa: black box problem
• Vị trí trong bài: Đoạn D, dòng 2-4
• Giải thích: “Many high-performing deep neural networks function as ‘black boxes’ – generating accurate predictions while providing limited insight into the causal mechanisms underlying those predictions.” Đây chính là vấn đề không thể hiểu được cách AI đưa ra dự đoán, khớp với đáp án B.

Câu 32: A

• Dạng câu hỏi: Matching Features
• Giải thích: “The difficulty of obtaining enough training examples for rare species” được thảo luận trong đoạn E về data requirements: “Endangered species, by definition, exist in limited numbers, providing few training examples for supervised learning algorithms.”

Câu 37: SHAP and LIME

• Dạng câu hỏi: Short-answer Questions (NO MORE THAN THREE WORDS)
• Vị trí trong bài: Đoạn D, dòng 10-11
• Giải thích: “SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) represent two such approaches”. Đây là hai kỹ thuật XAI được đề cập rõ ràng.

Câu 39: structural inequalities / power dynamics

• Dạng câu hỏi: Short-answer Questions (NO MORE THAN THREE WORDS)
• Vị trí trong bài: Đoạn F, dòng 5-7
• Giải thích: “Critics argue that framing conservation challenges as technical problems solvable through better algorithms obscures the underlying structural inequalities and power dynamics”. Cả hai cụm từ đều có thể là đáp án đúng.

---

5. Từ Vựng Quan Trọng Theo Passage

Passage 1 – Essential Vocabulary

Từ vựng	Loại từ	Phiên âm	Nghĩa tiếng Việt	Ví dụ từ bài	Collocation
artificial intelligence	n	/ˌɑːtɪˈfɪʃəl ɪnˈtelɪdʒəns/	trí tuệ nhân tạo	The use of artificial intelligence in wildlife conservation	AI technology, AI system, AI application
endangered species	n	/ɪnˈdeɪndʒəd ˈspiːʃiːz/	loài có nguy cơ tuyệt chủng	monitoring endangered species often require substantial resources	protect endangered species, save endangered species
machine learning	n	/məˈʃiːn ˈlɜːnɪŋ/	học máy (học tự động)	machine learning algorithms can automatically identify species	machine learning algorithm, machine learning model
camera trap	n	/ˈkæmərə træp/	bẫy ảnh (máy ảnh tự động)	camera trap technology has proven successful	camera trap image, deploy camera trap
poaching	n	/ˈpəʊtʃɪŋ/	săn bắt trái phép	predicting poaching activities	illegal poaching, anti-poaching, poaching incident
acoustic monitoring	n	/əˈkuːstɪk ˈmɒnɪtərɪŋ/	giám sát âm thanh	Acoustic monitoring represents another frontier	acoustic sensor, acoustic analysis
pattern recognition	n	/ˈpætən ˌrekəɡˈnɪʃən/	nhận dạng mẫu	pattern recognition software analyzes recordings	pattern recognition system, pattern recognition technology
predictive model	n	/prɪˈdɪktɪv ˈmɒdl/	mô hình dự đoán	predictive models can forecast threats	develop predictive model, use predictive model
biodiversity	n	/ˌbaɪəʊdaɪˈvɜːsəti/	đa dạng sinh học	protect our planet’s biodiversity	biodiversity loss, biodiversity conservation
conservation strategy	n	/ˌkɒnsəˈveɪʃən ˈstrætədʒi/	chiến lược bảo tồn	leading to revised conservation strategies	develop conservation strategy, implement conservation strategy
patrol route	n	/pəˈtrəʊl ruːt/	tuyến đường tuần tra	optimize their patrol routes	plan patrol route, follow patrol route
data privacy	n	/ˈdeɪtə ˈprɪvəsi/	quyền riêng tư dữ liệu	concerns about data privacy	data privacy issue, protect data privacy

Passage 2 – Essential Vocabulary

Từ vựng	Loại từ	Phiên âm	Nghĩa tiếng Việt	Ví dụ từ bài	Collocation
paradigm shift	n	/ˈpærədaɪm ʃɪft/	sự thay đổi mô hình/quan điểm cơ bản	catalyzed a paradigm shift	represent paradigm shift, experience paradigm shift
deep learning	n	/diːp ˈlɜːnɪŋ/	học sâu (loại AI)	deep learning algorithms	deep learning model, deep learning technique
convolutional neural network	n	/ˌkɒnvəˈluːʃənəl ˈnjʊərəl ˈnetwɜːk/	mạng nơ-ron tích chập	utilize convolutional neural networks	train convolutional neural network
anthropogenic	adj	/ˌænθrəpəˈdʒenɪk/	do con người gây ra	anthropogenic destruction caused by logging	anthropogenic impact, anthropogenic activity
marine conservation	n	/məˈriːn ˌkɒnsəˈveɪʃən/	bảo tồn biển	The application extends to marine conservation	marine conservation effort, marine conservation project
coral bleaching	n	/ˈkɒrəl ˈbliːtʃɪŋ/	tẩy trắng san hô	assess coral bleaching severity	coral bleaching event, prevent coral bleaching
habitat suitability	n	/ˈhæbɪtæt ˌsuːtəˈbɪləti/	tính phù hợp của môi trường sống	predicts habitat suitability for endangered species	assess habitat suitability, habitat suitability model
genomic conservation	n	/dʒiːˈnɒmɪk ˌkɒnsəˈveɪʃən/	bảo tồn gen	The realm of genomic conservation	genomic conservation effort, genomic conservation strategy
genetic diversity	n	/dʒəˈnetɪk daɪˈvɜːsəti/	đa dạng di truyền	analyze genetic diversity within populations	maintain genetic diversity, genetic diversity loss
breeding program	n	/ˈbriːdɪŋ ˈprəʊɡræm/	chương trình nhân giống	prioritized for breeding programs	captive breeding program, successful breeding program
algorithmic bias	n	/ˌælɡəˈrɪðmɪk ˈbaɪəs/	sự thiên vị thuật toán	Algorithmic bias poses a concern	address algorithmic bias, avoid algorithmic bias
computational resources	n	/ˌkɒmpjuˈteɪʃənəl rɪˈsɔːsɪz/	tài nguyên tính toán	computational resources required for AI	require computational resources, allocate computational resources
biodiversity hotspot	n	/ˌbaɪəʊdaɪˈvɜːsəti ˈhɒtspɒt/	điểm nóng đa dạng sinh học	grassroots organizations in biodiversity hotspots	protect biodiversity hotspot, identify biodiversity hotspot
open-source platform	n	/ˈəʊpən sɔːs ˈplætfɔːm/	nền tảng mã nguồn mở	develop open-source platforms	use open-source platform, create open-source platform
drone technology	n	/drəʊn tekˈnɒlədʒi/	công nghệ máy bay không người lái	combination with drone technology	deploy drone technology, drone technology application

Passage 3 – Essential Vocabulary

Từ vựng	Loại từ	Phiên âm	Nghĩa tiếng Việt	Ví dụ từ bài	Collocation
epistemological	adj	/ɪˌpɪstɪməˈlɒdʒɪkəl/	thuộc về nhận thức luận	fundamental epistemological transformation	epistemological question, epistemological implication
algorithmic turn	n	/ˌælɡəˈrɪðmɪk tɜːn/	sự chuyển đổi thuật toán	This algorithmic turn necessitates examination	represent algorithmic turn, experience algorithmic turn
neural network	n	/ˈnjʊərəl ˈnetwɜːk/	mạng nơ-ron	deployment of neural networks	train neural network, deep neural network
reinforcement learning	n	/ˌriːɪnˈfɔːsmənt ˈlɜːnɪŋ/	học tăng cường	The application of reinforcement learning	reinforcement learning agent, reinforcement learning algorithm
adaptive management	n	/əˈdæptɪv ˈmænɪdʒmənt/	quản lý thích ứng	adaptive management scenarios	adaptive management strategy, adaptive management approach
explainable AI	n	/ɪkˈspleɪnəbəl eɪ aɪ/	AI có thể giải thích được	employing explainable AI techniques	explainable AI method, develop explainable AI
transfer learning	n	/trænsˈfɜː ˈlɜːnɪŋ/	học chuyển giao	Transfer learning offers a solution	apply transfer learning, transfer learning approach
technocratic governance	n	/ˌteknəˈkrætɪk ˈɡʌvənəns/	quản trị kỹ trị	questions about technocratic governance	technocratic governance model, technocratic governance approach
indigenous wisdom	n	/ɪnˈdɪdʒənəs ˈwɪzdəm/	trí tuệ bản địa	marginalization of indigenous wisdom	respect indigenous wisdom, incorporate indigenous wisdom
algorithmic accountability	n	/ˌælɡəˈrɪðmɪk əˌkaʊntəˈbɪləti/	trách nhiệm giải trình thuật toán	Algorithmic accountability emerges as critical	ensure algorithmic accountability, lack algorithmic accountability
carbon footprint	n	/ˈkɑːbən ˈfʊtprɪnt/	dấu chân carbon	carbon footprint of artificial intelligence	reduce carbon footprint, measure carbon footprint
federated learning	n	/ˈfedəreɪtɪd ˈlɜːnɪŋ/	học liên kết	Federated learning approaches enable training	federated learning system, federated learning framework
multi-agent system	n	/ˈmʌlti ˈeɪdʒənt ˈsɪstəm/	hệ thống đa tác nhân	Emerging paradigms such as multi-agent systems	develop multi-agent system, multi-agent system approach
hybrid system	n	/ˈhaɪbrɪd ˈsɪstəm/	hệ thống lai/kết hợp	development of hybrid systems	create hybrid system, hybrid system design
ecological integrity	n	/ˌiːkəˈlɒdʒɪkəl ɪnˈteɡrəti/	tính toàn vẹn sinh thái	prioritize ecological integrity	maintain ecological integrity, protect ecological integrity
distributional shift	n	/ˌdɪstrɪˈbjuːʃənəl ʃɪft/	sự dịch chuyển phân phối	concerns about distributional shift	distributional shift problem, detect distributional shift
causal mechanism	n	/ˈkɔːzəl ˈmekənɪzəm/	cơ chế nhân quả	limited insight into causal mechanisms	understand causal mechanism, identify causal mechanism
morphological characteristic	n	/ˌmɔːfəˈlɒdʒɪkəl ˌkærəktəˈrɪstɪk/	đặc điểm hình thái	focused on morphological characteristics	analyze morphological characteristic, morphological characteristic analysis

---

Kết Bài

Chủ đề “How is AI being used in environmental conservation?” không chỉ phản ánh xu hướng công nghệ hiện đại mà còn là một trong những topic xuất hiện ngày càng thường xuyên trong IELTS Reading. Việc nắm vững chủ đề này giúp bạn tự tin hơn không chỉ trong phòng thi mà còn trong việc cập nhật kiến thức về các vấn đề toàn cầu đương đại.

Bộ đề thi mẫu trên đã cung cấp đầy đủ 3 passages với độ khó tăng dần từ Easy (Band 5.0-6.5), Medium (Band 6.0-7.5) đến Hard (Band 7.0-9.0). Mỗi passage được thiết kế theo đúng chuẩn Cambridge IELTS, với độ dài, cấu trúc câu hỏi và mức độ học thuật phù hợp với từng level. Tổng cộng 40 câu hỏi với 7 dạng khác nhau giúp bạn làm quen với đa dạng question types thường gặp trong thi thật.

Phần đáp án chi tiết không chỉ đưa ra đáp án đúng mà còn giải thích rõ ràng vị trí thông tin trong bài, cách paraphrase và lý do tại sao các đáp án khác không chính xác. Đây là phần vô cùng quan trọng giúp bạn tự đánh giá và học hỏi từ những sai lầm.

Bảng từ vựng theo từng passage với đầy đủ phiên âm, nghĩa, ví dụ và collocations sẽ giúp bạn mở rộng vốn từ học thuật, đặc biệt là từ vựng liên quan đến công nghệ, môi trường và bảo tồn – những chủ đề “hot” trong IELTS hiện nay.

Hãy sử dụng bộ đề này như một bài thi thật: làm trong điều kiện có thời gian (60 phút), sau đó đối chiếu đáp án và phân tích kỹ những câu sai. Đừng quên ghi chú lại từ vựng mới và luyện tập sử dụng chúng trong văn cảnh khác. Chúc bạn ôn tập hiệu quả và đạt band điểm mong muốn trong kỳ thi IELTS sắp tới!