AI GOVERNANCE

Let’s get down to business and meet your instructor Anna Bethke. She’s the Director of Data Science and Ethical AI at Included AI. Now she’s partnering with ELVTR to teach the blueprint to AI governance!

• Instructor introduction
• Course objectives & flow
• Final project overview
• Brief overview, key moments, and rapid growth of the AI industry
• Q & A
  

With great AI power, comes greater AI responsibility. It has the ability to create solutions, protect wildlife, and invade users’ personal privacy – all at once. It’s important to consider the unintended consequences of AI to assess how responsible you’re being and to find areas where you could be more effective.

• Evolution of AI and its impact on society
• Potential positive and negative consequences of AI
• Case Study: Positive & negative examples of AI from Meta, Google, WeWalk, etc.
• Workshop: Analyze an AI system to explore user & societal impact

Assignment 1: AI System Impacts
Select an AI system, and fill out a pre-formatted slide outlining its potential positive and negative consequences.

AI, like a newborn, has the potential to become anything. As responsible adults, it can be a challenge to keep up with how fast AI is growing. It can quickly head down the wrong path and jump into an even worse path. We’re going to cover diverse frameworks and tools to help you guide your AI intelligently.

• Overview of ethical AI
• Key principles in ethical AI
• Development of AI standards
• Case Study: Global initiatives & collaborative standards (NIST & GDPR) 

Assignment 2: Applicable AI Governance themes
Identify relevant AI Governance themes for the chosen AI system in assignment 1 with concise 1-2 bullet points explaining their applicability.

Ethical prevention is better than an algorithmic cure. So, how can we make AI systems better for everyone? Now that the groundwork is in place, let’s build upon our risk assessment skills and stay ahead of the tech game.

• Ethical considerations in AI development
• Identifying, assessing & mitigating AI risks
• Implementing ethical AI practices
• Demo: Deon
• Workshop: Mitigating risks in AI systems

Assignment 3: Risk Mitigation
Propose risk mitigation techniques for potential negative consequences identified in assignment one, as part of final project building blocks.

Oops…something went wrong! AI is guaranteed to make mistakes. Turn your “oops” into feedback loops while taking accountability. Some mistakes are minor, other mistakes can majorly impact your users and business. Today you’ll practice handling undesirable content.

• Regulatory & organizational oversight
• AI accountability, transparency & legal frameworks
• Accountability across levels: leaders & teams
• Workshop: The process of reporting inappropriate & inaccurate content

Humans have biases, and so do our creations. Intentional or not bias also exists within AI. How can we proactively reduce them? Let’s find out!

• Algorithmic bias and fairness
• Types of bias & fairness definitions
• Case Study: Impact of fairness decisions (COMPAS recidivism)
• Demo: Google's What-If Tool
• Guest Speaker: IBM AI Fairness 360 correcting bias in ML models

Assignment #4: Fairness definition understanding
Match fairness names with definitions and potential use cases using tools like the What-If tool, IBM’s 360 toolkit, Aquitas, and lecture notes.

Transparency builds trust. When it comes to AI every user wants to trust the brands they engage with. But what does that actually look like in practice? Get ready to practice creating model cards, like nutrition labels for AI systems, for some hands-on experience.

• Transparency overview
• AI transparency methods
• Model cards - nutrition labels for AI systems
• Workshop: Complete a model card for an AI system

Assignment 5: Model Card Creation
Fill out a model card for the system of your final project AI system following the templates provided.

Laws are being established to protect users’ data. The proposed Artificial Intelligence Act and other location-based regulations have set standards on how we collect, process, store, and leverage personal data. You’ll learn how these systems impact AI and how to create better systems that champion privacy.

• Data privacy regulations and compliance
• Differential privacy, how it works, and how to implement it
• Balancing innovation and privacy concerns
• What can users do to protect their data
• Workshop: Create strategies to prevent sensitive information leaks in AI systems

Ethics are subjective. With an international community able to access the same technologies, we may all have a different perspective on how ethical they are. How can you keep the whole world satisfied with your AI governance? Learn how to overcome these challenges when building AI systems.

• Comparative analysis of AI governance models
• Challenges and opportunities in global AI cooperation
• Cross-border data governance
• Guest Speaker
  

Assignment 6: Global AI Governance
Fill in a table of countries and AI governance themes, specifying inclusions and exclusions in these models.

Your AI might be ethical, but how can you prove this to your users? After all, the way you handle user data is the way you treat your customers. Dive into how big businesses develop ethical AI practices, avoid “ethics washing”, and navigate pitfalls.

• Corporate ethics in AI development
• Dangers of “ethics washing”
• Balancing profit & social impact
• Workshop: CSR in the age of AI
  

We’ve covered how AI policies within big businesses are developed. It’s time to zoom out and see how these policies can be created and put into play at a larger level.

• Creating effective AI policies
• Policy implementation challenges & solutions
• Leading organizations & ways to get involved
• Guest Speaker - Data, society, policy development & advocacy

Assignment 7: AI Governance Pitch
Craft 1-2 slides outlining a proposed AI policy, covering governance themes, regulatory points, etc.

A quick Google search of “generative AI problems” will show you what needs to be prevented. Generated pictures, videos, and texts can sometimes be hard to differentiate from the desired output. Other times they are directly problematic. Regulation has its place, and you’ll find out where to put it in the process.

• Generative AI overview
• Particular AI governance concerns
• Generative AI used ethically - how?
• Case Study: Mainstream generative AI (social media + news)
• Guest Speaker 

Regulation is racing against time. As AI launches itself further into innovation, governance is trying to keep up. Today is your chance to practice creating governance on a new technology to prepare yourself for the future.

• AI & autonomous systems
• The future of AI governance & challenges
• Case Study: Regulating AI in emerging industries
• Workshop: Creating governance on a new tech 

Unlock and open new career opportunities. Learn how to get started with the job hunt and discover the different available roles. Present your final project, say goodbye to your peers, and say hello to them on LinkedIn!

• Types of careers in that ecosystem
• AI governance work ranges
• Advancing your skills
• Demo - job boards
• Final project: 5-minute presentation of model card (for selected students)
  

Final Project: Present your 10 slides showcasing how to apply AI governance principles to a real-world scenario.