IEB · Session 09 · Exam Preparation

Algorithmic Decision-Making
& Trust

Exploring how organizations delegate authority to algorithms and the challenges of ensuring alignment, trust, and accountability.

Principal-Agent Theory in AI

Dowding & Taylor (2024) apply the Principal-Agent (P-A) framework to understand the delegatory relationship between humans and algorithms. In this model, the Principal (the firm) delegates tasks to an Agent (the AI).

Key Agency Problems (Dowding & Taylor, 2024)

  • Information Asymmetry: The Agent (AI) possesses complex decision logic that the Principal cannot easily verify.
  • Misalignment: The AI might optimize for a different goal (e.g., short-term accuracy) than the Principal's long-term strategic objective.
  • Agency Costs: The costs of monitoring the AI, creating evaluation benchmarks, and the "residual loss" from imperfect decisions.
PRINCIPAL Delegation of Decision Power Asymmetric Output/Feedback ALGORITHMIC AGENT
Fig. 1 — The P-A Relationship in AI. The "Black Box" nature of algorithms creates a severe information asymmetry that humans must manage through monitoring (Dowding & Taylor, 2024).

Trust in Algorithmic Systems

Trust is a prerequisite for delegation. Session 09 Course Materials differentiate between different bases of trust when dealing with opaque algorithmic systems.

🛡️

Institution-Based Trust

Trusting the system because it's governed by regulations, audits, and professional standards (Dowding & Taylor, 2024).

🔬

Explainability (XAI)

The technical ability to make the "Black Box" transparent. If we understand the *why*, we can trust the *what*.

🤝

Relational Trust

Trust built over time through consistent, reliable performance of the algorithm in practice.

Algorithmic Internalities

Dowding & Taylor (2024) introduce Algorithmic Internalities: the harms or suboptimal outcomes borne by the users/principals themselves due to their choice of algorithm.

Internalities vs. Externalities

While society worries about bias (Externalities), the firm worries about Internalities:

  • Loss of Human Capability: De-skilling of workers who become over-reliant on AI.
  • Strategic Narrowing: The algorithm might optimize for past patterns, missing radical future innovations.
  • Financial Loss: Bad predictions leading to failed R&D or poor resource allocation.

Novo Nordisk — Governing the Agent

Applying Session 09 logic to the OpenAI partnership:

  • Agency Costs at Novo: The cost isn't just the OpenAI subscription. It's the cost of Novo's scientists verifying every lead the AI generates. If they don't verify, the Agency Risk (bad drug leads) is too high.
  • Institutional Trust: Novo must rely on Institutional Trust. They need clear governance frameworks for how clinical data is used and how AI results are validated against biological ground truth.
  • Black Box Risk: In pharmaceutical research, a "Black Box" prediction isn't enough for the FDA. Novo needs Explainability to transform AI-output into scientific knowledge.

Likely Oral Exam Questions