Core Exam Guide
Artificial Intelligence Policy
Overall Guidelines
- The exam will be in class, paper and pen (no pencils, please - they smudge!). No computers will be used or needed. Write legibly.1
- The exam will have two sections - IDs and short answer.
- An “ID”, or identification, question, simply gives you a term and asks you to define it and identify its relevance for the study of AI and AI Policy.
- A short answer is like a short essay - nothing terribly long, but should be structured, potentially with short paragraphs. You could also be given a news article or data graphic about AI and be asked to interpret and situate it, saying why it is important to the broader discussion of AI.
- I do not imagine the exam will take the entire class period, but you will have the entire class period to complete it.
Definitions and Key Concepts
Philosophy of AI
- Foundations
- Intelligence
- Consciousness
- Sentience
- “Common Sense”
- Autonomous Systems
- Simulation
- Emergence
- Game of Life
- AI Design & Assessment
- Strong AI
- Weak AI
- Turing Test
- Chinese Room Argument
- Embodiment
- Agency
- Responsibility
- Role of Language in Thought & Intelligence
Technology of AI
- Hardware
- Turing Machine
- Basic parts of a computer (processor, memory, etc)
- Serial vs. Parallel Processing
- CPUs vs GPUs (and other, more specialized processing)
- Moore’s Law
- Silicon - design vs. fabrication
- “Die shrinks”
- Chip fabrication requirements
- Software
- Low vs. High Level programming languages
- Modeling
- Machine Learning
- Large Language Models
- Tokens
- Transformers
- Omitted Data/Algorithmic Bias
- Out of Sample Predictions/Long-Tail Problems
- Overfitting
Business of AI
- Building AI Systems
- Infrastructure
- Bubbles
- Products
- B2B vs Consumer
- Disruption
- Moats
- Startup tradeoffs: Profit vs Market Share
- Revenue streams
- AI product possibilities
- Market positions of current major tech & AI companies
- “Hyperscalers”
- Implications
- Data Access
- Energy
- Labor Replacement
- Materials & supply chains
Example Short Answer
- Given what we know about AI, how would you advise a firm looking to adopt an AI system to support its employees? What kinds of concerns should the firm have, and how might it ameliorate them?
- What is the difference between “Weak AI” and “Strong AI”?
- What is the Turing Test, and is it a sufficient “test” of AI?
- What would be an example of algorithmic bias in the world, why might it have arisen, and what is a proposal for managing it?
Footnotes
I’ll try to decipher everything, but ultimately, if I can’t read it, it can’t earn points.↩︎