Continued exploration of AI Epistemology and Practice

Continued exploration of AI Epistemology and Practice

September 12, 2023
AI Epistemology and Practice

Today’s Objective

Continuing from the previous week, Week 4 deepens the exploration of the epistemological foundation of AI. Students will delve into the complexities of scientific paradigms and how they shape AI research methodologies. Building on Pickering's insights and exploring Kuhn's perspective, the class aims to enhance understanding of how scientific knowledge is constructed, evolved, and challenged in the context of AI.

Lecture Notes

<insert slides link>

Key Concepts

  • Epistemology (Review): The study of knowledge, especially concerning its methods, validity, and scope (from Pickering's "The Mangle of Practice"). It answers the question: “why do we know what we know?”
  • Practice (Review): The application or use of ideas, beliefs, or methods, as opposed to theories about such application or use.
  • Scientific Paradigms: The set of practices, theories, and assumptions that constitute a scientific discipline (from Kuhn's "The Structure of Scientific Revolutions").

To Read (Before Class)



Student Facilitation Reading

"Epistemic Cultures: How the Sciences Make Knowledge" by Karin Knorr Cetina

(Intro & Chapter 1)

Summary: Knorr Cetina’s work explores how different scientific disciplines create and validate knowledge. It’s relevant to the examination of AI's epistemological foundation, delving into the various paradigms that shape AI research methodologies.

Critical Questions:

  1. How do different scientific cultures influence the development and understanding of AI?
  2. How can we reconcile conflicting epistemic paradigms in AI research?
  3. What role does community or collaboration play in the epistemology of AI?
  4. How might the epistemic cultures of AI influence its application in different fields?
  5. What are the implications of the "constructed" nature of scientific knowledge in AI?

To Watch

In Class Assignment

Storyboarding the Mangle of AI Technologies


To scrutinize the intricate web of relationships and actions that form around AI technologies. The aim is to gain an understanding of how AI technologies contribute to or challenge the production and validation of knowledge. These should be detailed an extensive user stories. You’ll end up using 20 to 40 sticky notes.


Students will form small groups and utilize storyboarding—a UX methodology traditionally used for visualizing user experiences—to examine the "mangle" associated with a chosen AI technology. Storyboarding is a UX design technique. It visually represents the steps or actions that an individual takes in interacting with a system. In this case, the "system" is broad and includes not just the AI technology but also the ecosystem of human and non-human actors around it. Each "frame" of the storyboard will represent a significant interaction or event that reveals how knowledge is created or validated.

Activity Details

  1. Step 1: Choose a Persona
    • Select a persona that will guide your perspective throughout this activity. Examples include a legislator, product manager, game designer, executive, or environmentalist.
    • Discuss as a group why you chose this persona and what you think their primary concerns or focus areas would be.
  2. Step 2: Select a AI Technology
    • Choose from the provided list of cutting-edge AI technologies. Make sure it aligns with the interests or focus areas of your chosen persona.
    • Companies to select from
      1. LLaMa, GPT-4 & Language Models
        • Examples: OpenAI's GPT-4, Google's LaMDA
        • Key Considerations: Natural language understanding, ethical usage, content creation
      2. Healthcare Predictive Analytics
        • Examples: Tempus Labs, Health Catalyst
        • Key Considerations: Patient data privacy, medical ethics, actionable insights
      3. Emotion Recognition Systems
        • Examples: Affectiva, Kairos
        • Key Considerations: Emotional intelligence, consent, potential misuse
      4. Conversational Agents in Mental Health
        • Examples: Woebot, Wysa
        • Key Considerations: Patient care, medical oversight, therapy limitations
      5. Generative Adversarial Networks (GANs) in Art Creation
        • Examples: Artbreeder, DeepArt
        • Key Considerations: Copyright issues, artistic agency, economic impact
      6. Human-Robot Interaction (HRI) Systems
        • Examples: SoftBank Robotics' Pepper, Boston Dynamics' Spot
        • Key Considerations: Social norms, autonomy, ethics in design
      7. Synthetic Biology and AI
        • Examples: Zymergen, Ginkgo Bioworks
        • Key Considerations: Bioethics, gene editing, sustainability
      8. Supply Chain Optimization
        • Examples: Llamasoft, ClearMetal
        • Key Considerations: Economic scalability, human job displacement, operational efficiency
      9. AI in Climate Modeling
        • Examples: ClimateAI, IBM's GRAF
        • Key Considerations: Environmental impact, data accuracy, policy implications
      10. Edge AI for Localized Computing
        • Examples: Nvidia's Jetson platform, Google Coral
        • Key Considerations: Data privacy, computational efficiency, IoT integration
      11. Explainable AI (XAI) Systems
        • Examples: IBM's AI Explainability 360, Fiddler AI
        • Key Considerations: Algorithmic transparency, accountability, ethical considerations
  3. Step 3: Identify Key Actors and Elements
    • Brainstorm and jot down both human and non-human actors involved. Consider technologies, laws, social norms, and economic factors.
  4. Step 4: Create Storyboard Using Sticky Notes
    • On desk or whiteboard, use sticky notes to create a storyboard that narrates a typical interaction or process involving your chosen technology.
    • The storyboard should depict how the key actors and elements interact, influence, and shape the technology and its surrounding ecosystem.
    • You should have 20 to 40 sticky notes to capture the process in detail. If you don’t know an interaction or the way a technology work, look it up and just take a best guess.
  5. Step 5: Identify Knowledge-Creation Moments
    • Use a special marker or symbol next to interactions that involve the creation or validation of knowledge.
    • Examples:
      • How does Amazon's recommendation engine process user data to suggest products?
      • How does IBM Watson in healthcare interpret medical records?
  6. Step 6: Annotate the Power Dynamics
    • Make notes on how different actors gain or lose power in this network, affecting the creation or validation of knowledge.
    • Examples:
      • How does Facebook's newsfeed algorithm prioritize certain kinds of knowledge (news, updates) over others?
      • How does a legislator's influence affect the regulation and thus the operational framework of AI in autonomous vehicles?
  7. Step 7: Annotate with three fitting quotes that explain your map from The Mangle of Practice
    1. Find three detailed direct quotes that help explain the phenomena you’re seeing in your map.

Questions for Post-Activity Discussion

Answer the following questions, write them down, and post to Teams (this is also your Journal assignment for this week)

  1. Epistemic Shifts: How does AI redefine traditional ways of knowledge creation?
  2. Comparison with Past Technologies: How does AI's role in knowledge creation compare with that of the printing press or the internet?
  3. Power and Control: Who holds the reins when it comes to dictating what is considered valid knowledge within this AI ecosystem?

To Do

Read today’s materials before class & take notes on your reading
Come to class and participate in the readings and activities
‼️ Write a one page (~300 words) journal prompt on: How will AI express the power to create and validate knowledge? What are the similarities here between AI and other technologies? Be sure to use the readings as background and include at least one quote or key passage from the reading.
Turn in all your assignments on Friday on eLearning