Coursera Plus Annual is 40% Off — Ends July 6.

Class Central is learner-supported. When you buy through links on our site, we may earn an affiliate commission.

Udacity

Generative AI Data Ethics

via Udacity

Go to class Write review

Overview

This course provides a comprehensive exploration of ethical considerations in AI development and deployment. Beginning with an introduction to responsible AI frameworks, you will grasp the significance of addressing bias in generative models and effective mitigation strategies. Key lessons cover the explainability of models, data licensing, and the auditing of data provenance. The course further delves into societal impacts, environmental sustainability, and security vulnerabilities related to data leaks. You will design ethical risk mitigation plans and explore human-in-the-loop systems. Ultimately, you will conduct a technical ethical audit of a generative AI tool, equipping them with practical skills to navigate complex ethical landscapes in AI.

Syllabus

  • Course Introduction
    • Get started with Generative AI Data Ethics: review prerequisites, tools, environment setup, and meet your instructor for this course.
  • Understanding Responsible AI Frameworks
    • Explore why responsible AI frameworks are needed, how they guide risk management and decision-making, and examine real examples for building ethical, accountable AI systems.
  • Understanding Bias in Generative Models
    • Learn to spot and evaluate subtle bias in generative AI models by analyzing patterns in tone, framing, and outputs, and using systematic, evidence-based review methods for fairness.
  • Mitigating Bias in Generative Models
    • Learn to identify, measure, and reduce bias in generative AI outputs using counterfactual testing, bias signal metrics, and post-processing mitigation techniques for fairer models.
  • Understanding Generative Model Explainability (XAI)
    • Explore generative AI explainability by focusing on system behavior, input visibility, controlled variation, and analyzing patterns for oversight and responsible deployment.
  • Interpreting Generative Model Outputs (XAI)
    • Learn to interpret generative model outputs using XAI principles, test prompt sensitivity, apply quantitative analysis, and document behavioral risks for trustworthy, auditable AI deployment.
  • Understanding Data Licensing and Copyright
    • Learn to identify legal and ethical risks in AI data use, address copyright, fair use, licensing, data lineage, and use guardrails for responsible and compliant AI development.
  • Auditing Data Provenance and Licensing
    • Learn to audit AI training data by tracing its provenance, interpreting licenses, identifying risks, and ensuring ethical, legal, and compliant data use in generative AI projects.
  • Understanding Societal and Economic Impact
    • Explore AI's societal and economic impacts, from changing work to trust in information and responsible design at scale.
  • Developing Ethical Risk Mitigation Plans
    • Learn to analyze AI ethical audits, prioritize and mitigate risks, and create actionable, accountable plans to ensure safer, more responsible AI systems.
  • Understanding Environmental Impact and Sustainability
    • Learn how effective AI system design impacts environmental sustainability, shifting focus from training to inference, with shared responsibility and principles of efficient, Green AI development.
  • Implementing Efficiency and Cost Optimization
    • Learn to optimize generative AI systems by analyzing tradeoffs in cost, efficiency, and quality. Apply data-driven methods for practical, scalable, and sustainable deployments.
  • Understanding Data Leaks and Security Vulnerabilities
    • Learn how generative AI systems leak data via oversharing, context bleed, and prompt injection, and explore strategies like guardrails to prevent unauthorized exposure of sensitive information.
  • Preventing and Responding to Data Leaks
    • Learn to build a layered AI defense against data leaks using automated guardrails, human-in-the-loop review, and incident response plans for responsible, secure AI deployment.
  • Understanding Algorithmic Auditing
    • Learn the essentials of algorithmic auditing: structured evidence-based evaluation of AI behavior for risk, fairness, and governance using checklists, scenarios, and repeatable audit processes.
  • Performing a Technical Ethical Audit
    • Learn how to systematically audit generative AI models for ethical risks by detecting sensitive data leaks, scoring outputs, reviewing risks, and preparing defensible audit artifacts.
  • Understanding Human-in-the-Loop Systems
    • Explore human-in-the-loop systems: integrate human oversight at critical decision points in AI to ensure accountability, mitigate risks, and build responsible, trustworthy automation.
  • Designing Human-in-the-Loop Systems
    • Learn to design Human-in-the-Loop (HITL) AI systems by integrating human oversight, risk-based routing, feedback loops, and workflows for safer, accountable, and effective AI content generation.
  • Project: Conduct an Ethical Audit and Mitigation Plan for a Generative AI Tool
    • You will review a fine-tuned language model in Jupyter, evaluate its outputs, use explainability to assess bias, and deliver an Ethical Audit, Mitigation Plan, and Ethics Committee presentation.

Taught by

Kesha Williams

Reviews

Start your review of Generative AI Data Ethics

Go to class

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Sign up for free
Someone learning on their laptop while sitting on the floor.