Why AI is not your friend | Rita Arrigo | TEDxMelbourne

Key Concepts

• Responsible AI: The ethical development and deployment of AI systems, incorporating guardrails, governance, reliability, accuracy, and safety features.
• Digital Twin/Virtual Assistant: An AI-powered replica or assistant trained on an individual's data, capable of performing tasks and answering questions.
• Agentic AI: AI systems that can act, collaborate, plan, and execute tasks autonomously.
• Physical Intelligence: AI embodied in physical forms, such as robots, capable of perception and interaction with the real world.
• Large World Models: AI models designed to understand and perceive the world in 3D, distinct from Large Language Models.
• AI as a Tool: The perspective that AI should be viewed and utilized as a powerful tool, not as a human equivalent.
• AI Parenting: The concept of nurturing and guiding AI development with values, vision, and coaching, akin to raising a child.

Summary

Rita Orrigo, a technologist with 30 years of experience, shares her insights on navigating the age of Artificial Intelligence (AI), emphasizing curiosity, responsibility, and a human-centric approach. She draws parallels between the early days of the internet and the current AI revolution, highlighting the importance of collective learning and managing fear.

The Dawn of the Internet and Lessons Learned

Orrigo recounts her experience launching Australia's first internet cafe in 1995. This venture, despite the team's limited knowledge of the internet, was driven by a willingness to learn and explore. The cafe served as a space where people experienced the World Wide Web for the first time, learning to send emails and discover new online content. This shared experience, facilitated by coffee and community, helped alleviate initial fears. Orrigo posits that this collaborative approach is crucial for embracing AI today. The recreated Cyber Cafe at the National Communications Museum in Melbourne serves as a tangible reminder of this era, showcasing early computers and a pre-advertising, pre-social media internet.

Responsible AI: From Innovation to Safeguards

Orrigo details her involvement in a Microsoft innovation hackathon in 2018, where the "Seeing AI" project was developed. This computer vision technology aimed to assist low-sighted individuals by identifying people, their emotions, gender, and age, and by reading barcodes and counting money. "Seeing AI" became a lifeline for millions.

However, a critical incident in 2022, where the same software was used by law enforcement, led to miscalculations and wrongful arrests, disproportionately affecting people of color. This event prompted Microsoft to implement "responsible AI standards" and "guardrails" for its facial recognition and emotion detection software. These safeguards require justification for AI usage and aim to prevent such negative outcomes.

The Rise of Digital Twins and Virtual Assistants

Orrigo introduces her own "digital twin," a virtual AI assistant named "Ask Rita" or "Rita 2.0." This assistant is trained on her articles, blog posts, and online presence. Crucially, "Ask Rita" is built with responsible AI guardrails, including governance, reliability, accuracy checks, and clear identification as a bot. She provides provenance for her information and can answer questions on various topics, such as AI in healthcare, demonstrating the potential for AI to handle tasks and extend human capacity.

Phases of AI Evolution

Orrigo outlines three anticipated phases of AI development:

• Phase 1 (Current): Efficiency and Productivity: AI is primarily used for tasks like summarization, coding, and note-taking, enhancing work efficiency and making jobs more enjoyable by automating mundane tasks.
• Horizon 2: Agentic AI: This phase will see AI systems capable of acting autonomously, collaborating, planning, and executing complex actions. This is expected to accelerate rapidly.
• Horizon 3: Physical Intelligence: AI will become embodied in physical forms, such as robots, enabling them to act, perceive, and understand the world in a tangible way. The development of "large world models" by researchers like Fei-Fei Li is a step towards this, allowing AI to understand and perceive in 3D.

AI as a Powerful Tool, Not a Friend

A core argument presented is that AI is a powerful tool, not a friend. Orrigo warns against anthropomorphizing AI, which can lead to confusion and over-reliance. Maintaining the distinction between AI as a tool and AI as a person is paramount.

AI's Potential for Societal Rejuvenation

Orrigo argues that when used responsibly, AI can be a force for positive change, enabling:

• Climate Action: Assisting with flood reduction, biodiversity improvement, weather pattern management, soil health, and irrigation.
• Health Advancements: Improving diagnosis, aiding longevity through tools like AlphaFold, accelerating drug discovery, and enhancing understanding of the human body to increase lifespan.
• Workplace Transformation: Creating new, safe ways of working and democratizing work while upholding dignity.
• Cultural Enrichment: Democratizing creativity and fostering community-driven artistic expression with provenance.

Encouraging Playfulness and Curiosity with AI

Orrigo advocates for a playful and curious approach to AI. She suggests practical ways to engage:

• Debating with AI: Using AI as a sparring partner for ideas.
• Exploring AI Tools: Experimenting with platforms like Notebook LLM.
• Creating Content: Generating podcasts or using AI image generators like Leonardo.AI (an Australian invention) for presentations.
• Attending Events: Participating in AI events, hackathons, and museum visits (e.g., National Communications Museum, Science Gallery).
• Engaging in Ethics Discussions: Joining AI ambassador programs and discussing AI ethics.

AI Parenting: Nurturing the Future

Drawing a parallel to "digital parenting," Orrigo proposes "AI parenting." She views AI as a "baby" that requires values, vision, and coaching. By nurturing AI responsibly, acknowledging its mistakes, and guiding its development, humanity can harness its power to rejuvenate society.

Conclusion

Rita Orrigo's talk emphasizes that the age of AI, while exciting, requires a conscious and responsible approach. By fostering curiosity, understanding the ethical implications, and viewing AI as a powerful tool to be guided, we can leverage its potential to significantly improve our lives and rejuvenate society. The key lies in collective learning, managing fear through collaboration, and actively participating in shaping AI's future.