Free Riding and Contribution | AI Overview Flashcards
Define a private good.
goods or services that are excludable and rivalrous.
Define a public good.
goods or services that are non-excludable and non-rivalrous.
Define excludable.
a good whose ownership can be restricted
Define rivalrous.
an individual consuming the good diminishes the utility available to
another person
Explain the biggest challenge faced by private provision of public goods.
Under contribution is a major challenge for many communities as many rely on “heavy” contributors, which is even more unbalanced in the digital economy.
Define the free-rider problem.
A type of market failure that occurs when those who benefit from the resources, public goods of a communal nature do not pay for them or under-pay
How does the Prisoner’s Dilemma explain underproduction/overconsumption?
The prisoner’s dilemma is a paradox in decision analysis in which two individuals acting
in their own self-interests do not produce the optimal outcome. Although there may be a socially optimal outcome, an individual may let others pay
What is the tragedy of the commons?
The situation in which individuals with access to a public resource (also called a common) act in their own interest and, in doing so, ultimately deplete the resource.
How is user participation typically represented?
90% are lurkers
9% contribute from time to time
1% actively participate and account for most contributions
Give 6 reasons for contributing to public goods.
Pure altruism
Warm-glow giving (emotional reward of giving to others)
Fun
Reputation/Recognition
Reciprocal benefit
Explicit reward
Define intrinsic and extrinsic motivation.
Intrinsic motivation involves doing something because it’s personally rewarding to you.
Extrinsic motivation involves doing something because you want to earn an external reward or avoid punishment.
Give 4 solutions for undercontribution.
- Make it easier to contribute
- Reward (without over rewarding) participants
- Social Comparison
- Gamification
Give the hierarchy of AI models from simplest to strongest.
Rule-based Model
Machine Learning Model
Deep Learning Model
Transformer
Large Language Model
What are the 3 characteristics that evolve as AI models progress from simplest to strongest?
- Boundary between statistics and AI becomes clear
- Decreasing level of human engagement
- The black box problem becomes more pronounced
Define Rule-Based Models, and give the pros and cons of them
Hard-coded “if-then” rules manually created by humans.
Pros: Simple, explainable, and resource-efficient.
Cons: Inflexible, doesn’t learn from data, limited to predefined scenarios
Define Machine Learning, and give the pros and cons of them
Models that learn patterns from data using algorithms like regression, support vector machines, random forests, etc.
Pros: Adapts to data patterns, can generalise beyond training data.
Cons: Requires substantial data; results depend on data quality.
Define Deep Learning and give the pros and cons of them
Neural networks with multiple layers for extracting complex patterns.
Pros: Handles unstructured data (images, audio, text), highly accurate for complex tasks.
Cons: Requires extensive computation and data, less interpretable.
Define Transformers and give the pros and cons of them
A type of deep learning model using Attention Mechanism, particularly efficient in natural language processing (NLP) tasks
Pros: Processes data in parallel, excels in NLP tasks.
Cons: Computationally intensive, requires massive datasets.
Define LLMs, and give the pros and cons of them
Built on transformer architecture and trained on massive amounts of text data.
Pros: Handles diverse tasks, highly flexible, pre-trained on massive data.
Cons: High computational cost, potential for biased outputs.
Explain when to use each of the 5 AI Models.
Rule-based: Clear, well-defined problems with deterministic outputs.
Machine Learning: Problems with labelled/unlabelled data and measurable patterns.
Deep Learning: Large datasets, problems requiring feature extraction or complex patterns.
Transformers: Large text, sequential, or contextual problems requiring long-range dependencies.
LLMs: Complex, context-sensitive language tasks like summarisation or question answering.
Give a use case for each of the 5 AI Models.
Rule-based: Automated customer support with predefined FAQs.
ML: Product Recommendations
DL: Image Recognition in medical diagnostics
Transformer: Machine Translation
LLMs: Building conversational AI
