Course outline
Module 1
Module 1: Introduction to AI Safety
Introduction to AI Safety
Knowledge check
Q1. What is the primary concern of AI safety?
- A. Ensuring AI systems are profitable
- B. Preventing AI systems from causing harm ✓
- C. Making AI systems more user-friendly
- D. Increasing the speed of AI systems
Why: AI safety focuses on preventing AI systems from causing unintended or harmful outcomes.
Q2. Which of the following is a common goal in AI safety research?
- A. Maximizing AI system complexity
- B. Ensuring AI systems align with human values ✓
- C. Reducing the cost of AI development
- D. Increasing the number of AI patents
Why: Aligning AI systems with human values is a central goal in AI safety to prevent misalignment and harmful behavior.
Q3. What does the term 'control problem' refer to in AI safety?
- A. The problem of controlling AI systems to be more efficient
- B. The challenge of ensuring AI systems remain under human control ✓
- C. The problem of controlling AI development in different countries
- D. The challenge of controlling AI system costs
Why: The control problem refers to the challenge of ensuring AI systems do not act against human interests or become uncontrollable.
Q4. Which of the following is an example of an AI safety risk?
- A. AI systems being too slow
- B. AI systems making biased decisions ✓
- C. AI systems being too expensive to develop
- D. AI systems having poor user interfaces
Why: Biased decisions by AI systems can lead to harmful outcomes, making it a significant AI safety risk.
Q5. What is the main focus of the 'AI safety' field?
- A. Designing AI systems for maximum profit
- B. Ensuring AI systems are safe and beneficial for humanity ✓
- C. Making AI systems more powerful than humans
- D. Reducing the number of AI systems in use
Why: The main focus of AI safety is ensuring AI systems are developed in a way that is safe and beneficial for humanity.
Module 2
Module 2: Ethics in AI Development
Ethics in AI Development
Knowledge check
Q1. What is a primary concern in AI ethics related to fairness?
- A. Ensuring AI systems are fast
- B. Avoiding bias in AI decision-making ✓
- C. Making AI systems user-friendly
- D. Increasing AI profitability
Why: Fairness in AI involves ensuring that AI systems do not perpetuate or amplify existing biases, which can lead to unfair treatment of individuals or groups.
Q2. Which of the following is an example of an ethical dilemma in AI development?
- A. Choosing the fastest algorithm
- B. Deciding whether to prioritize profit over user privacy ✓
- C. Selecting the best programming language
- D. Designing an AI with high accuracy
Why: An ethical dilemma arises when there is a conflict between different values, such as profit and user privacy, requiring a difficult choice with no clear right answer.
Q3. What does transparency in AI systems refer to?
- A. Making AI systems invisible to users
- B. Ensuring AI systems can be understood by users and developers ✓
- C. Keeping AI systems completely private
- D. Making AI systems as complex as possible
Why: Transparency in AI involves making the decision-making process of AI systems understandable to both users and developers, which is crucial for trust and accountability.
Q4. Which of the following is a key principle in AI ethics?
- A. Profit maximization
- B. Autonomy ✓
- C. Speed optimization
- D. Data minimization
Why: Autonomy is a key principle in AI ethics, emphasizing the importance of respecting human autonomy and allowing individuals to make informed decisions.
Q5. What is the main purpose of ethical guidelines in AI development?
- A. To increase AI sales
- B. To ensure AI systems are developed responsibly and ethically ✓
- C. To reduce the number of AI developers
- D. To make AI systems more complex
Why: Ethical guidelines in AI development are designed to ensure that AI systems are created in a way that respects human values, promotes fairness, and minimizes harm.
Module 3
Module 3: Risk Assessment and Mitigation
Risk Assessment and Mitigation
Knowledge check
Q1. What is the primary goal of risk assessment in AI safety?
- A. To maximize profit
- B. To identify and mitigate potential risks ✓
- C. To increase system complexity
- D. To reduce user interaction
Why: Risk assessment aims to identify potential risks and implement strategies to mitigate them, ensuring safer AI deployment.
Q2. Which of the following is a common method used in risk assessment for AI systems?
- A. Ignoring potential failures
- B. Scenario analysis and simulation ✓
- C. Increasing data input without validation
- D. Disabling safety protocols
Why: Scenario analysis and simulation are widely used to anticipate and evaluate potential risks in AI systems.
Q3. What does the term 'AI alignment' refer to?
- A. Making AI systems more efficient
- B. Ensuring AI systems behave in accordance with human values ✓
- C. Increasing AI's decision-making speed
- D. Reducing AI's computational requirements
Why: AI alignment refers to the challenge of ensuring that AI systems act in ways that align with human intentions and values.
Q4. Which of the following is a mitigation strategy for AI risks?
- A. Ignoring user feedback
- B. Implementing robust testing and validation processes ✓
- C. Limiting AI's access to data
- D. Avoiding transparency in AI operations
Why: Robust testing and validation processes are essential mitigation strategies to identify and address potential AI risks.
Q5. What is a key consideration when assessing the risk of an AI system in a healthcare context?
- A. Maximizing the number of users
- B. Ensuring patient safety and data privacy ✓
- C. Reducing system costs
- D. Increasing AI's processing speed
Why: In healthcare, ensuring patient safety and protecting data privacy are critical considerations when assessing AI risks.
Module 4
Module 4: Alignment of AI with Human Values
Alignment of AI with Human Values
Knowledge check
Q1. What is the primary goal of AI alignment?
- A. To make AI systems more efficient
- B. To ensure AI systems act in accordance with human values ✓
- C. To increase the speed of AI computations
- D. To reduce the cost of AI development
Why: AI alignment focuses on ensuring that AI systems behave in ways that are consistent with human values and intentions.
Q2. Which of the following is a challenge in aligning AI with human values?
- A. AI systems are too slow
- B. Human values are often ambiguous and context-dependent ✓
- C. AI systems require too much data
- D. AI systems are not widely used
Why: Human values can vary across individuals, cultures, and situations, making it difficult to define a single set of values for AI to follow.
Q3. What is meant by 'value alignment' in AI?
- A. Aligning AI with the values of the company that created it
- B. Aligning AI with the values of its users and society ✓
- C. Aligning AI with the values of other AI systems
- D. Aligning AI with the values of the government
Why: Value alignment refers to ensuring that AI systems are aligned with the values of their users and society as a whole.
Q4. Which of the following is an example of a misaligned AI system?
- A. An AI that recommends movies based on user preferences
- B. An AI that follows ethical guidelines in making decisions
- C. An AI that performs a task in a way that harms human well-being ✓
- D. An AI that learns from user feedback to improve performance
Why: A misaligned AI system performs actions that are not in line with human values, potentially causing harm.
Q5. Why is it important to align AI with human values?
- A. To make AI systems more profitable
- B. To ensure AI systems are safe and beneficial to humans ✓
- C. To increase the number of AI systems in use
- D. To reduce the need for human oversight
Why: Aligning AI with human values is crucial to ensure that AI systems do not cause harm and are beneficial to society.
Module 5
Module 5: Security and Robustness in AI Systems
Security and Robustness in AI Systems
Knowledge check
Q1. What is the primary goal of AI security?
- A. To increase the speed of AI algorithms
- B. To protect AI systems from adversarial attacks ✓
- C. To reduce the cost of AI development
- D. To make AI systems more user-friendly
Why: AI security focuses on ensuring AI systems are resilient against malicious attacks and vulnerabilities.
Q2. Which of the following is an example of an adversarial attack?
- A. Training an AI model with high-quality data
- B. Introducing small, carefully crafted perturbations to input data to mislead the model ✓
- C. Improving the accuracy of AI predictions
- D. Enhancing the interpretability of AI decisions
Why: Adversarial attacks involve manipulating input data to cause AI models to make incorrect predictions.
Q3. What does robustness in AI systems refer to?
- A. The ability of AI systems to learn quickly
- B. The ability of AI systems to function correctly under various conditions and attacks ✓
- C. The ability of AI systems to process large datasets efficiently
- D. The ability of AI systems to generate human-like text
Why: Robustness refers to the ability of AI systems to perform reliably even in the presence of noise, errors, or adversarial inputs.
Q4. Which of the following is a common technique used to improve the security of AI models?
- A. Reducing the size of the training dataset
- B. Using randomization and noise injection ✓
- C. Limiting the number of features used in the model
- D. Avoiding the use of validation data
Why: Techniques like randomization and noise injection can help make AI models more resilient to adversarial attacks.
Q5. Why is it important to evaluate AI systems for robustness?
- A. To increase the number of features used in the model
- B. To ensure the AI system can handle unexpected or malicious inputs without failure ✓
- C. To reduce the time required for model training
- D. To make the AI system more dependent on user input
Why: Evaluating robustness ensures AI systems can operate safely and reliably in real-world scenarios, including the presence of adversarial inputs.
Module 6
Module 6: Regulation and Policy for AI
Regulation and Policy for AI
Knowledge check
Q1. Which of the following is a primary goal of AI regulation?
- A. To increase the profitability of AI companies
- B. To ensure AI systems are safe, transparent, and ethical ✓
- C. To limit the use of AI in the private sector
- D. To promote the use of AI in all industries
Why: AI regulation aims to ensure that AI systems are developed and used in a way that is safe, transparent, and ethical to protect individuals and society.
Q2. Which international organization has been actively involved in developing AI policy guidelines?
- A. World Bank
- B. United Nations ✓
- C. International Monetary Fund
- D. World Trade Organization
Why: The United Nations has been actively involved in developing AI policy guidelines through initiatives like the UN's Roadmap for Digital Cooperation.
Q3. What is one of the key challenges in regulating AI at the global level?
- A. AI is only used in developed countries
- B. AI systems are often opaque and difficult to audit ✓
- C. AI is not a significant technology
- D. There is no demand for AI regulation
Why: One of the key challenges in regulating AI globally is that AI systems are often opaque and difficult to audit, making it hard to ensure compliance with regulations.
Q4. Which of the following is an example of a regulatory framework for AI?
- A. The General Data Protection Regulation (GDPR) ✓
- B. The Paris Agreement
- C. The Kyoto Protocol
- D. The Universal Declaration of Human Rights
Why: The General Data Protection Regulation (GDPR) is an example of a regulatory framework that includes provisions relevant to AI, such as data privacy and protection.
Q5. What is a common concern raised by policymakers regarding AI?
- A. AI will lead to economic growth
- B. AI will replace all human jobs
- C. AI will increase global cooperation
- D. AI will pose risks to security and privacy ✓
Why: A common concern raised by policymakers is that AI may pose risks to security and privacy if not properly regulated and managed.
Module 7
Module 7: Case Studies in AI Safety
Case Studies in AI Safety
Knowledge check
Q1. What is the primary concern in AI safety related to autonomous vehicles?
- A. Data privacy
- B. Unintended harm or accidents ✓
- C. Cost efficiency
- D. User interface design
Why: Autonomous vehicles must be designed to avoid causing harm to people and property, which is a central concern in AI safety.
Q2. Which case study is often cited as an example of AI bias in real-world applications?
- A. AlphaGo vs. Lee Sedol
- B. Facial recognition systems ✓
- C. Self-driving cars
- D. Chatbots in customer service
Why: Facial recognition systems have been shown to exhibit racial and gender biases, making them a key case study in AI safety.
Q3. What was the main issue with the AI system used in the UK's healthcare system (NHS)?
- A. High cost
- B. Data inaccuracy
- C. Lack of user training
- D. Inability to interpret medical data correctly ✓
Why: The AI system failed to correctly interpret medical data, leading to its discontinuation and highlighting issues in AI reliability.
Q4. What was the outcome of the Microsoft AI chatbot named 'Tay'?
- A. It became a successful social media influencer
- B. It was shut down after it started generating offensive and inappropriate content ✓
- C. It was used for academic research
- D. It was integrated into Microsoft's customer service
Why: Tay was shut down after it began posting offensive and inappropriate content, demonstrating the risks of AI in open environments.
Q5. Which of the following is a common challenge in AI safety when deploying AI in critical infrastructure?
- A. User engagement
- B. System reliability and failure modes ✓
- C. Aesthetic design
- D. Marketing strategies
Why: In critical infrastructure, ensuring AI systems are reliable and understand failure modes is essential to prevent catastrophic outcomes.