Quiz: AI and Machine Learning Fundamentals¶
Test your understanding of artificial intelligence, machine learning, and their applications to investor relations with these questions.
1. What distinguishes machine learning from traditional software programming?¶
- Machine learning systems improve performance through exposure to data rather than explicit programming
- Machine learning requires less computational power
- Machine learning can only work with numerical data
- Machine learning programs are always more accurate
Show Answer
The correct answer is A. Machine learning systems improve performance through exposure to data rather than following explicit programmed rules. Traditional software executes predefined logic, while ML systems learn patterns from data to make predictions or decisions. Option B is incorrect—ML often requires significant compute. Option C is wrong—ML works with text, images, audio, and other data types. Option D is false—ML accuracy depends on data quality and problem complexity.
Concept Tested: Machine Learning Basics
Bloom's Level: Understand
2. In supervised learning, what role does labeled training data play?¶
- It provides examples showing the relationship between inputs and correct outputs
- It replaces the need for any human involvement
- It guarantees 100% accuracy on new data
- It eliminates the need for model testing
Show Answer
The correct answer is A. In supervised learning, labeled training data provides examples showing the relationship between inputs (features) and correct outputs (labels), allowing the model to learn patterns for making predictions on new, unseen data. Option B is incorrect—humans label data and validate results. Option C is unrealistic—no model achieves perfect accuracy. Option D is wrong—testing on held-out data is essential for validation.
Concept Tested: Supervised Data Models
Bloom's Level: Understand
3. What is the primary purpose of unsupervised clustering algorithms?¶
- To predict future stock prices
- To label training data automatically
- To group similar data points together without predefined categories
- To replace human decision-making entirely
Show Answer
The correct answer is C. Unsupervised clustering algorithms group similar data points together based on patterns in the data, without predefined categories or labels. In IR, this could segment investors by behavior patterns or group similar earnings call questions by topic. Option A describes predictive modeling, not clustering. Option B reverses cause and effect—unsupervised learning doesn't require labels. Option D overstates AI capabilities—clustering supports, not replaces, human judgment.
Concept Tested: Unsupervised Clustering
Bloom's Level: Understand
4. What are Large Language Models (LLMs)?¶
- Databases storing large amounts of language data
- Translation tools for converting between programming languages
- AI systems trained on vast text datasets enabling human-like language understanding and generation
- Spell-checking software for long documents
Show Answer
The correct answer is C. Large Language Models are AI systems trained on vast text datasets that enable human-like language understanding and generation. Examples include GPT-4, Claude, and Gemini. These models can write, summarize, analyze, and respond to text in contextually appropriate ways. Option A describes databases, not AI models. Option B describes compilers or transpilers. Option D describes simple text tools, not sophisticated AI systems.
Concept Tested: Large Language Models
Bloom's Level: Remember
5. In prompt engineering, what is the purpose of providing "few-shot examples" to an LLM?¶
- To reduce the model's processing time
- To permanently retrain the model
- To reduce the cost of API calls
- To show the model the desired output format and style through concrete examples
Show Answer
The correct answer is D. Few-shot examples demonstrate the desired output format and style through concrete examples within the prompt, helping the model understand expectations without retraining. For instance, showing 2-3 examples of how to summarize earnings calls guides the model to produce similar summaries. Option A is incorrect—examples may increase prompt length. Option C mischaracterizes prompting—it doesn't retrain models. Option D may be backwards—longer prompts with examples can increase costs.
Concept Tested: Prompt Engineering Skills
Bloom's Level: Apply
6. What distinguishes agentic AI systems from traditional AI applications?¶
- Agentic systems require more data storage
- Agentic systems are always more expensive
- Agentic systems can only process text data
- Agentic systems operate autonomously, making decisions and taking actions without continuous human intervention
Show Answer
The correct answer is D. Agentic AI systems operate autonomously, perceiving their environment, making decisions, and taking actions to achieve goals without continuous human intervention. Unlike traditional AI that responds to specific inputs, agentic systems can plan multi-step workflows and adapt strategies. Option A is irrelevant to the distinction. Option C may or may not be true depending on implementation. Option D is incorrect—agentic systems can be multimodal.
Concept Tested: Agentic AI Systems
Bloom's Level: Understand
7. Your company wants to use an LLM to analyze earnings call transcripts. What is a critical consideration for enterprise LLM deployment?¶
- LLMs work perfectly without any configuration
- Implementing governance controls for data privacy, accuracy validation, and usage monitoring
- LLMs can replace all human IR functions immediately
- Enterprise LLMs don't require security measures
Show Answer
The correct answer is B. Enterprise LLM usage requires implementing governance controls including data privacy protections (preventing sensitive data exposure), accuracy validation (human review of outputs), usage monitoring (tracking what data is processed), and compliance frameworks. Option A is dangerously naive—LLMs require careful configuration. Option C overstates capabilities—LLMs augment, not replace, human expertise. Option D ignores critical security requirements for enterprise data.
Concept Tested: Enterprise LLM Usage
Bloom's Level: Apply
8. What is "reinforcement learning" in the context of AI systems?¶
- Repeatedly showing the same training data
- An ML approach where systems learn optimal strategies through trial, feedback, and rewards
- Strengthening computer hardware for AI workloads
- A technique for compressing large datasets
Show Answer
The correct answer is B. Reinforcement learning is an ML approach where systems learn optimal strategies through trial and error, receiving feedback in the form of rewards or penalties. The system explores actions, observes outcomes, and adjusts strategy to maximize cumulative rewards. In IR, this could optimize investor engagement timing or communication strategies. Option A describes data augmentation or repeated training, not RL. Option C relates to infrastructure, not algorithms. Option D describes data compression techniques.
Concept Tested: Reinforcement IR Learning
Bloom's Level: Remember
9. What are generative AI tools?¶
- Tools that only analyze existing data
- Manual writing assistants
- Software applications that create new content based on learned patterns
- Traditional search engines
Show Answer
The correct answer is C. Generative AI tools are software applications that create new content (text, images, code, audio) based on patterns learned from training data. Examples include ChatGPT for text generation, DALL-E for images, and GitHub Copilot for code. Option A describes analytical AI, not generative. Option C describes word processors or grammar checkers. Option D describes information retrieval systems, not content generation.
Concept Tested: Generative AI Tools
Bloom's Level: Remember
10. What role do training datasets play in machine learning model development?¶
- They are only needed during initial setup
- They replace the need for model validation
- They guarantee models will work on any future data
- They provide historical examples that teach ML systems patterns for making predictions
Show Answer
The correct answer is D. Training datasets provide historical examples that teach ML systems patterns, enabling them to make predictions on new data. Quality and representativeness of training data directly impact model performance. Option A is incomplete—models may need retraining with new data. Option C is backwards—validation data (separate from training) is essential. Option D overstates—training on past data doesn't guarantee performance on all future scenarios, especially if conditions change.
Concept Tested: Model Training Datasets
Bloom's Level: Understand
Quiz Statistics¶
- Total Questions: 10
- Bloom's Taxonomy Distribution:
- Remember: 3 questions (30%)
- Understand: 5 questions (50%)
- Apply: 2 questions (20%)
- Answer Distribution:
- A: 2 questions (20%)
- B: 2 questions (20%)
- C: 3 questions (30%)
- D: 3 questions (30%)
- Concepts Covered: 10 of 12 chapter concepts (83%)
- Estimated Completion Time: 15-20 minutes
Next Steps¶
After completing this quiz:
- Review the Chapter Summary to reinforce AI/ML concepts
- Work through the Chapter Exercises for hands-on practice
- Proceed to Chapter 6: AI-Powered Content Creation