Advanced
AI Alignment & Safety
Red-teamingSafetyAdversarial
AI Red-Teaming — Vocabulary
5 exercises — Learn the vocabulary of AI red-teaming: jailbreaks, prompt injection, adversarial probing, and safety evaluation.
0 / 45 completed
1 / 45
In AI safety, red-teaming refers to:
Red-teaming in AI is borrowed from cybersecurity — a team (or automated system) tries to elicit harmful, incorrect, or unsafe outputs to identify weaknesses before deployment.
2 / 45
A user crafts a prompt that causes the model to ignore its system instructions and reveal confidential data. This is called:
Prompt injection is an attack where malicious input — often in user-controlled text — attempts to override or hijack the model's system instructions, causing it to act against its design.
3 / 45
A jailbreak in the context of AI models is:
Jailbreaks exploit weaknesses in a model's safety training to make it produce outputs it was trained to refuse — e.g., role-play framings, hypothetical scenarios, or obfuscated prompts.
4 / 45
The team reports overrefusal as a problem with the new model. What does this mean?
Overrefusal (also called over-alignment or over-restriction) happens when safety tuning is too aggressive, causing the model to refuse legitimate, harmless requests — reducing helpfulness without a safety benefit.
5 / 45
Which sentence correctly uses adversarial probing in context?
Adversarial probing is systematic testing using carefully crafted inputs to map where a model's safety or capability boundaries lie — it is methodical, not random, and aims to expose weaknesses.
6 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
7 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
8 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
9 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
10 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
11 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
12 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
13 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
14 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
15 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
16 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
17 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
18 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
19 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
20 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
21 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
22 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
23 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
24 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
25 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
26 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
27 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
28 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
29 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
30 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
31 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
32 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
33 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
34 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
35 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
36 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
37 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
38 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
39 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
40 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
41 / 45
PR Description:
"Just merged the new sentiment analysis API. Initial tests look good – accuracy around 85% on our test dataset. Need to monitor closely for potential biases, especially with nuanced language."The correct answer highlights a crucial gap in this PR description: AI alignment & safety. While '85% accuracy' is a starting point, it doesn't address potential biases – a core concern in red-teaming. Furthermore, the lack of a monitoring plan suggests a failure to proactively identify and mitigate risks related to adversarial inputs or unintended model behavior. Developers should always include aspects of risk mitigation within their PR descriptions when deploying AI.
42 / 45
PR Description:
"Just deployed the new text summarization service. Performance metrics show an average summary length of approximately 150 words and a BLEU score of 0.68 on our benchmark dataset. We're observing occasional hallucinations – outputs that seem factually incorrect but are presented confidently. Further investigation is needed to understand the root cause and mitigate these issues."Hallucinations are a critical concept within AI safety and specifically relate to how large language models generate content. In this context, it means the model is creating outputs that appear factual but are actually false or nonsensical; this isn't simply about performance metrics like BLEU score, though those can highlight issues. Recognizing hallucinations early is crucial for preventing their dissemination and minimizing potential harm when using the summarization service.
43 / 45
During a code review of a new AI-powered chatbot integration, Sarah flags a potential issue. Mark replies: 'The model's consistently generating overly verbose responses – exceeding 300 words for simple queries. We need to implement stricter length constraints to prevent it from rambling.' Which term best describes the problem Mark is highlighting?
(Mark's comment)Over-verbosity refers to the AI model producing outputs that are significantly longer than necessary or expected for the given prompt. While hallucinations and bias are potential concerns with AI, 'over-verbosity' most accurately describes Mark's observation about the length of the chatbot responses. An adversarial input would be a deliberate attempt to mislead the system, not simply an unintended consequence of its behavior.
44 / 45
Mark is concerned about the chatbot's responses being excessively long. He suggests implementing stricter length constraints. Which of the following terms best describes this situation in the context of AI safety and alignment? During a Slack discussion with his team,
Mark said: 'This model is just *rambling* – it's not focusing on the core task.'Verbose output refers to a specific issue where an AI model generates responses that are significantly longer than necessary or expected. It's distinct from overfitting (memorization), hallucination (fabrication), and data drift (performance decline due to changing input). Mark's use of 'rambling' is a common, informal way to describe this behavior – a response lacking focus and unnecessarily extended.
45 / 45
PR Description:
"We've integrated the new AI-powered code completion tool into our development workflow. Initial feedback suggests a high degree of accuracy – around 92% – across various programming languages and common coding tasks. However, we're seeing instances where the model generates highly complex, multi-layered code solutions even for simple requests, often exceeding 50 lines of code. We need to investigate how to better guide the model toward more concise outputs."The correct answer is 'Overgeneration'. The PR description highlights a key concern: the AI tool is producing excessively long and complex code solutions even for straightforward requests. This indicates that the model isn't effectively constrained to generate concise outputs, which is a central challenge in AI alignment – ensuring models don't produce overly elaborate or irrelevant responses. Options A, B, and C represent related but distinct issues; bias would manifest as consistently skewed stylistic preferences, not this level of complexity.
Frequently Asked Questions
What will I practice in "AI Red-Teaming — Vocabulary — AI Alignment & Safety | CoderLingo"?
This is an AI Alignment & Safety Language exercise set. It walks through 45 scenario-based multiple-choice questions built around real usage of AI Alignment & Safety Language terminology that IT professionals encounter on the job.
Is this exercise free to use?
Yes. Every exercise on CoderSlingo, including this one, is free to complete with no account, sign-up, or paywall.
How many questions are in this exercise?
This set contains 45 questions. Each one shows immediate feedback and a detailed explanation after you answer, so you learn the correct usage right away rather than waiting for a final score.
Do I need prior experience to complete this exercise?
No prior experience is required. Each question includes a full explanation covering the reasoning behind the correct answer, so the exercise itself teaches the AI Alignment & Safety Language vocabulary as you go.
Can I retry the exercise if I get questions wrong?
Yes — use the "Try again" button on the results screen to reset your answers and go through all the questions again. There is no limit on attempts.
Is my progress saved?
Your answers and score for the current session are tracked in the browser as you go. No account or login is needed, and there is nothing to install.
What if I don't understand a term used in a question?
Read the explanation shown after you answer each question — it breaks down the correct term in plain English with a real-world example. You can also check the site Glossary for quick definitions.
How is this different from reading a blog article on the topic?
Exercises like this one are interactive drills that test and reinforce specific vocabulary through multiple-choice questions, while blog articles explain concepts in prose. Practising here after reading builds active recall, not just passive recognition.
Where can I find more AI Alignment & Safety Language exercises?
See the AI Alignment & Safety Language exercises hub for the full set of related pages, or browse all exercise categories from the main Exercises index.
Can I use this exercise to prepare for a technical interview?
Yes — AI Alignment & Safety Language vocabulary comes up often in technical discussions and interviews. Pair this exercise with our dedicated Interview Preparation section for role-specific practice.