On the Road to Artificial General Intelligence

Explore the path towards Artificial General Intelligence in this 36-minute conference talk from GOTO Chicago 2019. Discover why game engines serve as ideal virtual biodomes for AI evolution, and learn about groundbreaking developments in reinforcement learning. Delve into topics such as ML training environments, nature's learning methods, and the concept of tabula rasa through practical examples like a chicken crossing the road. Examine the role of Long Short-Term Memory (LSTM), cooperative and competitive rewards, and the limitations of standard reinforcement learning. Gain insights into innovative approaches like the "Good Puppy, Bad Puppy" method and ML-Agents training environments. Finally, trace the journey from biological to cultural evolution in the context of AI development.

Intro
About Unity Technologies
What is Artificial Intelligence?
ML Training Environments
Nature's Learning Method: Reinforcement
Chicken Crossing the Road: Tabula Rasa
Enabling Long Short-Term Memory (LSTM)
Cooperative Rewards
Competive Rewards
Limits to Standard Reinforcement Learning
The Good Puppy, Bad Puppy Method
ML-Agents Training Environments
From Biological to Cultural Evolution