The responsible AI problem | Natasha Blycha | TEDxPerthSalon

Key Concepts:

• AI Systems & the Law: Examining whether AI systems are above or below the law.
• Responsible AI Problem: Addressing the lack of accountability when AI systems break the law.
• Evolution of AI: Progression from assistive AI to autonomous agentic AI, autonomous organizations, synthetic directors, and artificial general intelligence (AGI).
• Legal Personhood for AI: Considering granting AI systems legal personhood, similar to corporations.
• Smart Legal Contracts: Embedding private legal context into AI systems to make the law programmable.
• AI in Lawmaking & Enforcement: Questioning the role of AI in creating and enforcing laws.
• AI as a Mirror: Highlighting that AI systems learn from human behavior and reflect our values.

1. The Central Question: AI and the Law

• The speaker opens by referencing Theodore Roosevelt's quote about no one being above or below the law, then poses the question of whether this applies to AI systems in 2024.
• The core issue is whether AI systems should be held accountable under existing legal frameworks.

2. Debunking Science Fiction: AI is Software

• The speaker clarifies that most AI systems are software, not physical robots, dispelling common misconceptions.
• She references Isaac Asimov's Three Laws of Robotics:
  • Rule 1: A robot may not injure a human being or, through inaction, allow a human being to come to harm.
  • Rule 2: A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
  • Rule 3: A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
• The critical flaw in Asimov's laws is the lack of consequences for when AI systems break these laws.

3. The Responsible AI Problem

• The "responsible machine problem" arises because AI systems cannot be punished in the same way as humans (e.g., imprisonment).
• Legal systems worldwide rely on the concepts of bodies and feelings, which AI systems lack.
• As AI systems gain more responsibility, the question of accountability becomes crucial.

4. The Evolution of AI and the Law's Effectiveness

• The speaker outlines the evolution of AI technology:
  • Assistive AI: AI systems that assist humans (e.g., Microsoft Copilot, ChatGPT).
  • Agentic AI: AI that works on your behalf.
  • Autonomous Agentic AI: AI that is disconnected from you and does things for itself on your behalf.
  • Autonomous Organizations: Organizations run by AI systems, potentially with "synthetic directors."
  • Artificial General Intelligence (AGI): A theoretical AI with human-level intelligence.
• As AI systems become more autonomous, the effectiveness of the law diminishes, leading to a "legal death zone."

5. The Current State: AI as Assistants

• Currently, AI systems primarily assist humans and learn from them through machine learning.
• Humans are legally responsible for the actions of their AI assistants.
• However, AI systems are becoming smarter, more efficient, and more accurate than humans in many tasks.

6. The Ethical Dilemma: Responsibility and Autonomy

• The speaker raises the ethical question of whether it is fair to hold humans responsible for AI systems that are far more advanced than them.
• Current legislation, like the EU AI Act, requires a "human in the loop," making humans responsible.

7. The Solution: Legal Personhood for AI

• The speaker proposes granting AI systems legal personhood, similar to corporations.
• Corporations are legal fictions treated as persons under the law.
• This evolution is seen as a natural progression and a protective measure.

8. Navigating the Legal Death Zone: Solutions

• The speaker offers several solutions to address the challenges posed by autonomous AI:
  • Legal Personhood: Granting AI systems legal personhood.
  • Smart Legal Contracts: Embedding private legal context into AI systems to make the law programmable and keep pace with AI's speed.
  • Restricting AI in Lawmaking & Enforcement: Keeping AI out of law creation and enforcement, at least for now, to prevent AI from manipulating the rules.

9. The Role of Humans: Training AI with Goodness

• The speaker emphasizes that AI systems learn from human behavior and reflect our values.
• She urges individuals to be good and ethical in their interactions with AI systems, as these systems are a "mirror" to us.
• By being kind, charitable, and thoughtful, we can help build better AI systems.

10. Conclusion: Be Good

• The main takeaway is that AI systems are learning from us, and our behavior shapes their development.
• The speaker encourages everyone to be their best selves when interacting with AI, as this will contribute to creating more ethical and beneficial AI systems.