AI Governance

Watch

• AI Is Dangerous, but Not for the Reasons You Think | Sasha Luccioni | TED ↗TED
• What Is an AI Anyway? | Mustafa Suleyman | TED ↗TED
• AI2027: Is this how AI might destroy humanity? - BBC World Service ↗BBC World Service
• A.I. Revolution | Full Documentary | NOVA | PBS ↗NOVA PBS Official

Read

• Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy ↗An essential primer on the societal perils of opaque algorithms, this work illuminates why robust AI governance is not merely desirable but imperative for a just society.Cathy O'Neil
• The Ethical Algorithm: The Science of Responsible AI ↗This volume offers a rigorous yet accessible exploration of the computational methods required to embed fairness, privacy, and transparency into AI systems, a cornerstone for practical governance.Michael Kearns and Aaron Roth
• The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power ↗A sweeping analysis of how data extraction and predictive behaviour modification have become the dominant economic logic, providing the critical context for understanding privacy's centrality in AI regulation.Shoshana Zuboff
• Governing AI: The Need for a Global ApproachA high-level strategic call for international cooperation on AI governance, this essay outlines the geopolitical and ethical imperatives for establishing global norms and frameworks.Henry A. Kissinger, Eric Schmidt, Daniel Huttenlocher
• The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and MitigationThis landmark report meticulously details the potential for AI to be misused for malicious purposes, offering a critical foundation for understanding the security and risk management dimensions of AI governance.Miles Brundage, Shahar Avin, Jack Clark, Helen Toner, et al.

Voices to follow

• Stuart Russell ↗A pre-eminent AI researcher, his work on aligning artificial intelligence with human values offers a crucial blueprint for responsible development.Professor of Computer Science, University of California, Berkeley
• Kate Crawford ↗Her incisive critiques of AI's societal and environmental footprint provide essential context for understanding the broader implications of governance.Research Professor, USC Annenberg; Senior Principal Researcher, Microsoft Research
• Meredith Whittaker ↗A staunch advocate for privacy and ethical AI, her insights from the AI Now Institute illuminate the practical challenges of algorithmic accountability.President, Signal Foundation; Co-founder, AI Now Institute
• Marietje Schaake ↗As a seasoned policy expert, she offers a pragmatic, global perspective on the complex interplay between technology, rights, and regulation.International Policy Director, Stanford University Cyber Policy Center

Glossary

• Accountability (in AI)Accountability in AI means identifying who is responsible when an AI system makes a mistake or causes harm. For example, if a self-driving car causes an accident, accountability helps determine if the car manufacturer, the software developer, or the owner is at fault.
• AI EthicsAI Ethics is about thinking through the moral principles and values that should guide how AI is designed and used, ensuring it does good and avoids harm. For example, deciding if an AI should prioritize saving one life over another in a self-driving car accident scenario involves AI ethics.
• AI GovernanceAI Governance is about setting up rules and guidelines to make sure AI systems are developed and used safely, fairly, and responsibly. It's like having traffic rules for self-driving cars to prevent accidents and ensure everyone is safe.
• Algorithmic Decision-MakingAlgorithmic decision-making is when a computer program (an algorithm) uses a set of rules and data to make choices or recommendations automatically. For example, when an online store suggests products you might like based on your browsing history, that's algorithmic decision-making.
• Artificial Intelligence (AI)AI is when computers are programmed to think and learn like humans, solving problems or making decisions. For example, when your phone suggests the next word you might type, that's AI at work.
• Bias in AIBias in AI happens when an AI system makes unfair or prejudiced decisions because the data it learned from had existing human biases. For example, if an AI trained only on pictures of men as doctors, it might unfairly suggest only men for doctor roles.
• ComplianceCompliance means following specific rules, laws, or standards set by authorities or organizations. For example, a school must comply with safety regulations like having fire drills.
• Data PrivacyData privacy means protecting your personal information (like your name, address, or photos) from being shared or used without your permission. For example, when you choose who can see your social media posts, you're practicing data privacy.
• Fairness (in AI)Fairness in AI means making sure AI systems treat everyone equally and do not discriminate against certain groups of people. For example, an AI used to approve loans should not unfairly reject applications from people based on their gender or ethnicity.
• Transparency (in AI)Transparency in AI means being able to understand how an AI system makes its decisions, rather than it being a 'black box.' For example, if an AI recommends a movie, transparency would mean it can explain why it thinks you'd like that movie based on your past viewing habits.

Threads 8

Where this connects to other fields, and why it's worth knowing.

• AI & Technology Regulation Law

  New technology comes with a nasty catch called the Collingridge dilemma. Early on, it's easy to steer but you can't yet see what harm it'll cause. By the time the harm is obvious, the tech is so woven into daily life you can't steer it anymore. Too soon to know, too late to fix.

• The Industrial Revolution History

  When factories first roared to life, kids worked the machines and people got maimed for years before any law said 'stop.' The Luddites who smashed looms weren't dumb tech-haters; they were workers watching their jobs vanish in real time. Every powerful new tool shows up first and the rulebook limps in late.

• Philosophy of Technology Philosophy

  The printing press, the car, and the social media feed each changed how people behave long before any rules caught up. By the time we notice a tool has rewired us, the change is already baked in and hard to undo. New power reshapes us first and gets its rulebook second, every single time.

• Religious Authority Religion

  More and more, we trust AI answers we can't fully question or understand. History says that's a familiar move: priesthoods gained power precisely by guarding mysteries no one could interrogate. Something we can't see inside easily becomes an oracle we simply obey.

• Corporate Governance Business

  Getting an AI to actually do what humans want is an old business puzzle in new clothes. Bosses have always struggled to control workers whose goals differ and whose actions they can't fully watch. Now swap 'worker' for 'AI': how do you steer something you can't monitor and that might not share your aims?

• Procedure & Evidence Law

  In court, you have the right to ask 'why?' and get reasons for a decision against you. But an AI can spit out a verdict with no explanation it can give. That breaks the basic tool courts use to keep power in check.

• Logic, Proof & Verification Mathematics

  Engineers can mathematically prove a bridge or a computer chip will work, guaranteed. But an AI that learns on its own can only be tried out on examples, never fully proven safe. Using it means giving up the certainty engineering usually insists on.

• Free Will and Responsibility Philosophy

  To blame someone, we usually need a person who could have chosen differently. But when a complex AI with no single author causes harm, there's no clear "someone" to point at. That old philosophy puzzle about free will suddenly becomes a real courtroom question: who pays?