Please enable JavaScript.
Coggle requires JavaScript to display documents.
The epistemology of AI - Coggle Diagram
The epistemology of AI
Constructivism and social constructivism
AI knowledge depends on context
Knowledge is constructed through interaction
Knowledge is not purely objective
Reality is constructed by the cognizer
Learning involves reconstruction of patters
Offer possibilities but interprets them
Actively constructed noit passively received
Not purely objective access to reality
Logical positivism and empiricism
Knowledge is justified true belief
Knowledge must be publicly expressible
Verification through evidence
Mind as representation system
AI processes through data and logic
Knowledge is meanginful
Scientific methodology and hypothesis
Model of knowledge creation
AI mirrors cognition
AI systems are compared to logical agents
Explainability and Interpretability
Interpretability builds trust
Validations is complex and continuous
Hard to understand how decisions are made
Interpretability affetcs user trust
Lack of transparency in AI models
Knowledge may evolve as system learns
AI autonomy complicates reponsibility
"Black boxes"
Bias and Fairness in AI
Inequalities
Fairness is subjective
AI can learn from human data
technical solutions may not solve bias
AI systems can amplify inequalities
Ethical responsibility falls on design
Present in training data
Legitimacy of AI generated knowledge
Historical data encode discrimination