Disclaimer: this interview was generated by CoPilot
Joe Rogan: Ladies and gentlemen, welcome to the Joe Rogan Experience. Today, we’ve got a super special guest with us, Miles Brundage. He’s an AI policy researcher, formerly with OpenAI, and someone who’s deeply involved in the development and ethical considerations of artificial general intelligence. Welcome, Miles!
Miles Brundage: Thanks, Joe. It’s great to be here.
Joe Rogan: So, let’s dive right in. AGI—Artificial General Intelligence. It’s like the Holy Grail of AI, right? Can you explain to the folks what AGI is and how it’s different from the AI we have today?
Miles Brundage: Absolutely. AGI refers to an AI system that possesses the ability to understand, learn, and apply intelligence across a broad range of tasks at a level comparable to human intelligence. Unlike narrow AI, which is designed for specific tasks like playing chess or recommending movies, AGI would have the flexibility and generality of human cognition. For example, while a narrow AI can be a champion at chess, it wouldn’t understand how to drive a car, compose music, or diagnose medical conditions without significant retraining. AGI, on the other hand, would be able to transfer its learning across different domains seamlessly.
Joe Rogan: Wow, that’s mind-blowing. We’re talking about machines that can essentially think and reason like us. How close are we to achieving AGI?
Miles Brundage: That’s a tough question. Predictions vary widely among experts. Some believe we could see AGI within a few decades, while others think it might take much longer. For instance, projects like OpenAI’s GPT-4, which powers advanced conversational AI like ChatGPT, and DeepMind’s AlphaGo Zero, which mastered the game of Go without human data, show tremendous progress in specific areas. However, they still lack the generality we’d expect from an AGI. Moreover, creating an AGI involves solving complex problems in understanding context, common sense reasoning, and abstract thinking, which are areas where current AI systems still struggle.
Joe Rogan: What are some of the biggest challenges you’re facing in developing AGI?
Miles Brundage: One of the main challenges is ensuring that AGI systems are safe and aligned with human values. The potential power of AGI is enormous, so we need to make sure that these systems act in ways that are beneficial and not harmful. For example, aligning AGI involves techniques like reinforcement learning from human feedback (RLHF), where AI learns desired behaviours based on human preferences. An example in practice is how autonomous driving systems, like those developed by Waymo, improve their performance based on user feedback and real-world driving data. Additionally, governance challenges include creating international policies and cooperation frameworks, given that AGI development could have global implications.
Joe Rogan: Right, because we’ve all seen those sci-fi movies where AI goes rogue. How do you ensure that AGI will be a force for good and not something that could be dangerous?
Miles Brundage: There are several approaches to this. One is technical alignment, which involves designing AGI systems that inherently understand and follow human values. For instance, OpenAI has been researching AI alignment methods like iterated amplification, where the AI is trained to break down complex tasks into simpler subtasks that it can solve and verify. Another approach is through governance, which includes creating policies and frameworks to guide the development and deployment of AGI. An example of governance in action is the European Union’s AI Act, which seeks to regulate AI systems and ensure they are used responsibly. Collaboration across different sectors and countries is crucial to developing these safeguards.
Joe Rogan: What role did OpenAI play in this landscape?
Miles Brundage: OpenAI is dedicated to ensuring that artificial general intelligence benefits all of humanity. We focused on both technical research and policy development. For instance, projects like the OpenAI Gym, which provides a toolkit for developing and comparing reinforcement learning algorithms, and the Dota 2 AI, which showcased our research in complex, real-time strategy games, have made significant strides. By collaborating with other organisations and policymakers, we aimed to create a broad coalition that can address the multifaceted challenges posed by AGI. Though I’ve recently resigned from my role there, I still believe in their mission and continue to work on AI policy and safety from a broader perspective.
Joe Rogan: It sounds like a lot of responsibility. On a lighter note, do you ever worry about AGI taking over jobs or making humans obsolete?
Miles Brundage: It’s a valid concern. While AGI has the potential to revolutionise industries and improve our quality of life, it could also disrupt job markets. For example, AGI could automate complex tasks in fields like medical diagnostics, as seen with IBM Watson, which assists doctors by analysing medical data and providing treatment recommendations. That’s why it’s essential to consider how we can transition to an economy that integrates AGI in a way that benefits everyone, potentially creating new kinds of jobs and opportunities. For instance, there could be roles in managing and overseeing AI systems, which would require human intuition and ethical judgement.
Joe Rogan: That’s a hopeful perspective. Before we wrap up, what excites you the most about the future of AGI?
Miles Brundage: The potential for AGI to address some of humanity’s most pressing challenges excites me. Whether it’s advancing medical research, like predicting protein folding with DeepMind’s AlphaFold, tackling climate change by optimising renewable energy solutions, or improving education through personalised learning systems like those developed by Coursera, AGI could play a significant role in finding innovative solutions. It’s a thrilling and daunting prospect at the same time.
Joe Rogan: Thanks for joining us, Miles. This has been an enlightening conversation. Folks, keep an eye on the developments in the world of AGI—it’s going to be a wild ride!
Miles Brundage: Thanks, Joe. It was great talking to you.
Joe Rogan: Before we go, I want to share some recent AI news. Did you hear about the AI tools slashing film production costs by 90%? At TIFFCOM, industry leaders discussed how AI can dramatically reduce costs and open up new investment opportunities. It’s amazing how AI is transforming industries!
Miles Brundage: Yes, I did! It’s incredible to see AI making such a big impact. And speaking of AI in everyday life, Google Play Store might soon introduce a feature allowing users to ask questions about apps, with AI providing answers. This could really enhance the user experience.
Joe Rogan: That’s awesome! It’s great to see AI being used to make things more accessible and efficient. Thanks for sharing that, Miles.