Module 3 of 3

Running Them Responsibly

Data, hallucinations, cost, and what governance looks like when you're not a developer

This module assumes you already understand that LLM outputs are probabilistic and that confidence and correctness are independent; the question is what to do about that in practice. Covers the specific data handling policies of the major enterprise platforms, how to design proportionate verification workflows for different risk levels, and what minimum viable governance looks like for a non-technical team.

An Intro to LLM Chatbots
  • ~50 mins
  • 3 lessons
  • foundation
Get this course

What you'll learn

  • Describe what happens to data sent to an LLM chatbot and what the key differences between the major enterprise platform policies mean for your organisation
  • Design a verification workflow proportionate to task risk: which categories require human review, what a meaningful check looks like in practice, and how to avoid overhead that drives teams toward ungoverned shadow use
  • Define the minimum viable governance framework for an organisation rolling out LLM chatbots
  • Identify the vendor questions that must be answered before an organisational deployment

Lessons in the Module

1 Data, Privacy, and What Goes to the Model 20–25 mins
2 Hallucinations, Confidence, and Verification 20–25 mins
3 Governance Without Engineering 20–25 mins
Back to course overview