The Next Big Questions in AI Research with Andrew Ng | ASK MORE OF AI with Clara Shih

Listen in as Clara sits down with world-renowned AI researcher Andrew Ng, founder of DeepLearning.ai, Coursera, and Google Brain, as they discuss the next frontiers of AI research, the way kids engage in unsupervised learning, and how his lifelong interest in AI started with flying helicopters.
Everyone seems to be rushing into AI and asking questions later. We’ll give you the info you need to get this right: sforce.co/4486KzY
Want even more AI? We've got you covered! askmoreofai.com
#Salesforce #AskMoreofAI
Chapters:
0:00 Introduction
0:55 What were you researching in 2002 at Standford University?
2:27 Did you think you would see an AI passing "The Turning Test?"
3:41 What was it like to work with AI in the early 2000s?
8:57 How did you transition to deep learning and neural networks?
12:24 What are your thoughts on reinforcement learning coming back?
13:41 Discussing Chat GPT
16:20 What role do large companies, start-ups, and academia play in AI?
19:13 Discussing Salesforce Einstein 1 and Copilot
20:18 What are the most exciting questions in AI right now?
24:20 What is the role of creativity in AI?
26:12 What risks are you concerned about with AI?
29:20 What would it take to create an AI system smarter than humans?
31:12 Tell us about your voice clone experiment
33:00 How do we trust audio and video in an AI era?
37:30 How do you think your upbringing helped you?
39:40 How should we educate our kids about AI?
42:30 How do you stay current with everything?
45:10 Closing thoughts
Subscribe to Salesforce’s CZcams Channel: bit.ly/SalesforceSubscribe
Learn more about Salesforce: www.salesforce.com
Facebook: / salesforce
Twitter: / salesforce
Instagram: / salesforce
LinkedIn: / salesforce
About Salesforce: Salesforce is the customer company, helping businesses of every size and industry put the customer at the center since 1999. Salesforce helps companies connect with customers in a whole new way with a foundation of trust, harnessing AI + Data + CRM with Salesforce Customer 360, its complete portfolio of products that unites every team around the customer on one integrated, low code platform. Guided by core values and driven by a global community of Trailblazers transforming careers, companies and communities, Salesforce is more than a business - it’s a platform for change. Learn more at www.salesforce.com.

[@salesforce]

Everyone seems to be rushing into AI and asking questions later. We’ll give you the info you need to get this right: sforce.co/4486KzY

[@benyawarath]

Thank you both so much! I really enjoyed hearing Dr. Ng's thoughts about the AI regulatory directions in the US (from min 33.00), how important teachers are, and how we should prepare kids for the future.

[@HarpaAI]

🎯 Key Takeaways for quick navigation:
00:00 🧠 Andrew Ng's early career and research focus
- Andrew Ng's early research focused on reinforcement learning and robotics.
- He shifted to deep learning and perception research when faced with the challenge of getting robots to see.
- Organized a workshop titled "Towards Human-Level AI" to discuss the long-term future of AI.
02:29 🤖 Aspirations for human-level AI
- Andrew Ng was motivated by the idea of achieving human-level AI throughout his career.
- Initially believed that scaling up neural networks would lead to significant progress.
- Acknowledged that the path to achieving artificial general intelligence was more challenging than anticipated.
05:25 📊 Perceptions of AI progress
- Andrew Ng didn't perceive the AI winter as a setback because he saw continuous year-on-year progress in the field.
- Recognized that public perception of AI had its ups and downs, but the actual progress was consistent.
- Highlighted the importance of market timing in technology adoption.
08:54 🧠 Creativity and progress in technology
- Discussed the idea that creative acts in technology often result from incremental progress over time.
- Gave examples of how technological advancements may appear sudden to outsiders but are built on years of effort.
- Emphasized the importance of persistence and patience in achieving success.
11:53 🔄 Transition from reinforcement learning to deep learning
- Andrew Ng initially combined reinforcement learning with small neural networks for robotics.
- Explained his motivation for transitioning to deep learning, influenced by theories of intelligence.
- Highlighted the focus on unsupervised learning in the early days and its evolving importance.
15:52 🌐 Role of academia, large companies, and startups in AI research
- Discussed the significance of different organizations in advancing AI research.
- Acknowledged the distribution advantage of large companies in bringing technology to users.
- Highlighted the speed of decision-making and innovation in startups.
- Emphasized the continuous research contributions from academia and their importance.
20:55 🤖 AI Research Directions
- Andrew Ng discusses the most interesting questions in AI research today.
- Highlights areas of excitement and potential in AI research.
21:55 🧠 Unsupervised Learning and Vision Models
- Andrew Ng expresses his continued excitement about unsupervised learning.
- He mentions the challenges and opportunities in training large vision models.
22:24 🤖 AI Agents
- Andrew Ng discusses the challenges and potential of AI agents, especially in decision-making.
- He acknowledges that this area is still evolving.
23:27 🌐 Data-Centric AI and Edge Applications
- Andrew Ng emphasizes the importance of data-centric AI for practical applications.
- He predicts the growing significance of AI at the edge.
27:59 🌍 AI Risks and Extinction
- Andrew Ng discusses real AI risks such as bias, fairness, and inaccuracy.
- He considers the idea of AI leading to human extinction as overblown and unlikely.
29:32 🧠 Path to AI Smarter Than Humans
- Andrew Ng talks about the challenges in achieving artificial general intelligence (AGI) and its timeline.
- He reflects on the different paths to intelligence between digital and biological entities.
31:32 🗣️ Voice Cloning and Trust
- Andrew Ng shares his experience with voice cloning technology.
- He discusses the importance of trust in audio and video interactions in the age of AI.
33:04 📜 AI Regulation and Watermarking
- Andrew Ng expresses concerns about AI regulation and regulatory capture.
- He advocates for smarter regulation, including the use of watermarking to indicate AI-generated content.
37:39 🌏 Upbringing and Thinking Differently
- Andrew Ng reflects on how his upbringing in different places influenced his perspective and thinking.
- He highlights the importance of thinking differently and being comfortable with contrarian ideas.
39:39 📚 Education for an AI-Driven World
- Andrew Ng discusses the importance of lifelong learning in a rapidly changing world.
- He advocates for teaching everyone to code, given the increasing accessibility and importance of AI tools.
42:12 🖥️ Learning to Code for Everyone
- Andrew Ng advocates for a future where everyone learns to code, just like learning a first language.
- He believes that coding in languages like Python is crucial for telling computers what you want unambiguously.
- Lifelong learning is essential as technology evolves rapidly, and coding skills will empower individuals in an AI-driven world.
45:22 🧠 Key Takeaways from the Discussion
- Andrew Ng's interest in AI research started with flying helicopters.
- He sees significant potential in unsupervised learning, image processing, and edge computing.
- Those deeply involved in AI research never experienced an AI winter, as they continued to make remarkable discoveries in their field.
Made with HARPA AI

[@zepingluo694]

Thanks for posting this nice summary

[@th-shubham]

This guy is a gem 👏👏

[@miladkhademinori2709]

Exactly 💯

[@christine_california]

Is AI speaking for Clara?

[@raghavendrasooda5368]

We need to use AI to estimate poverty level of an individual or family that can be used by government to provide food, education, health care to poor people

[@darylallen2485]

3:20 - I'm not sure she's grasped that LLMs aren't the same as human level ai.