Artificial Intelligence System Safety (AISYS)


Course Description

Artificial Intelligence and autonomous systems have become pervasive in software applications. However, the implications for AI safety has created new concerns and risk factors especially for autonomous vehicles. This course provides a thorough survey of AI, machine learning, optimization, and autonomous vehicle techniques followed by safety and hazard analysis methods. Along the way, we will bring clarity to definitions, actual capabilities of AI systems, and the current state of data science. Case studies and real-world incidents will also provide learnings and insights to further advance the goal of AI safety.

Objectives: To provide a survey of current AI and autonomous vehicle techniques as well as discovering and mitigating their hazards.   This course will also apply existing software safety methods to artificial intelligence and establish why AI is just another tool in a suite of software functionalities.

Who Should Attend: Engineers, Program Managers and Senior Managers having responsibility for the development and acceptance of systems operating with AI/ML.

Prerequisites: Individuals should be familiar with the concepts and processes of System Safety and Software Safety.

Course
AISYS 26-1
AISYS 26-2
Dates
15-19 Sep 2025
16-20 Mar 2026

Course Outline


  1. Definitions and Concepts
  2. AI and ML Capabilities
    • Optimization and routing
    • Metaheuristics and tree search
    • Machine Learning
    • Deep Learning
    • Simulations
    • Reinforcement learning
    • Agents and Agentic Systems
  3. ML Explainability
    • Data analysis and preparation
    • Statistical methods
    • Scientific method versus data mining
    • Confusion matrices, ROC, and AUC
    • P-values and P-hacking
    • Sampling bias, outliers, and data rot
  4. Autonomous Systems
    • Defining autonomous vehicles
    • Levels of autonomy
    • Air and ground vehicle instrumentation
    • Air and ground vehicle automation methods
    • Ego, connected, modular, and end-to-end systems
    • Localization techniques
    • Object detection techniques
    • Environment mapping techniques
    • Human factors in autonomous vehicles
    • Generative AI Concerns
  5. Data Science Talent Strategy
    • Defining data science
    • State of data science as a discipline
    • Rubrics for hiring a data scientist
  6. AI Risk Assessment
    • Where does AI Safety fit in the life cycles, your organization, your safety programs
  7. AI Hazard Analysis
    • Data Safety Analysis
    • Functional Hazard Analysis (FHA)
    • Operational Domain and Operational Design Domain (ODD) Analysis
    • AI Development, Verification, Validation, and Testing
  8. AI System Safety (defined as loss of life or property, illness, environmental damage) and "AI Safety" (defined as illegal, immoral, unethical)
    • Data Safety management and metrics
    • Machine Learning (ML) process assurance
    • LLM assurance and
  9. Large Language Model (LLM), Safety Prompting, Test, and Metrics
  10. Operationalizing your own AI Policy, Standards, and Adoption
    • Safety Management Systems (SMS)
    • Organization opportunities

CEU: 3.2

Course Duration: 4.5 Days

Tuition: $3,300 (July 2025 - June 2026)

This program is open to all eligible individuals. The Aviation Safety and Security Program operates all of its programs and activities consistent with the University’s Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor.

Published on May 3rd, 2021Last updated on April 22nd, 2025