Keynote Lecture by Dr. Andrew Ng

Introduction

• Presenter: Dr. Andrew Ng
• Positions:
  • Managing General Partner of AI Fund
  • Founder of DeepLearning AI and Landing AI
  • Chairman and Co-Founder of Coursera
  • Adjunct Professor of Computer Science at Stanford
• Notable Past Work:
  • Started and led Google Brain Team
  • Director of Stanford AI Lab
• Educational Impact: Over 8 million people have taken AI classes from him.

AI as a General Purpose Technology

• Analogy: AI is likened to electricity in its versatility and applicability across different fields.
• Main Tools Discussed:
  • Supervised Learning: Good at labeling and mapping input to output.
  • Generative AI: New and exciting, enables creation of content.

Applications of Supervised Learning

• Examples:
  • Online advertising, spam detection, self-driving cars, ship route optimization, automated visual inspection in factories.
• Workflow:
  • Collect labeled data.
  • Train AI model with labeled data.
  • Deploy and run AI model via cloud services.

Evolution of AI Tools

• Last Decade: Growth in large-scale supervised learning with large datasets and compute.
• Current Decade: Rise of generative AI, exemplified by models like ChatGPT.
• Advancements:
  • Large language models predict next word or token, using supervised learning.

Generative AI as a Developer Tool

• Shift in Development:
  • Previously took 6-12 months to build AI systems.
  • Now can be done in hours or days with prompt-based AI.
• Code Example: Simple sentiment classifier using Python.

AI Opportunities and Future

• Current Value: Majority of financial value from supervised learning.
• Future Projections:
  • Supervised Learning will continue to grow.
  • Generative AI will grow significantly in the next 3 years.
• Potential for Startups: New opportunities for startups and existing companies to create value.

Adoption Challenges

• Concentration in Tech: AI mainly adopted in consumer software/internet.
• Long Tail Opportunity: Many smaller, valuable projects in other industries.
• Solutions:
  • Low-code/no-code tools to enable industry customization.

AI in the Economy

• AI Stack:
  • Hardware and Infrastructure: Capital intensive, few winners.
  • Developer Tools: Hyper-competitive, some mega-winners.
  • Applications: Opportunities for unique applications with less competition.

Building Startups

• AI Fund Strategy:
  • Validate ideas quickly.
  • Recruit CEOs early.
  • Build prototypes with customer validation.
• Case Study: Bearing AI, used to optimize ship fuel efficiency.

Risks and Social Impact

• Ethical Considerations: Avoid projects that are not beneficial to humanity.
• Bias and Fairness: AI systems improving but still have issues.
• Job Disruption: AI automating higher-wage jobs.
• AGI Concerns: AI as AGI is decades away, not seen as an extinction risk.

Conclusion

• AI as Opportunity: General-purpose technology creating diverse opportunities.
• Call to Action: Engage in building concrete AI use cases.
• Closing: Thank you for listening.