Explore the journey towards Artificial General Intelligence in this thought-provoking conference talk from GOTO Copenhagen 2018. Delve into the role of intelligence in biological evolution and learning, and discover why game engines serve as ideal virtual biodomes for AI evolution. Examine how the scale and speed of simulations are revolutionizing AI development, with insights from Danny Lange, VP of AI and ML at Unity Technologies. Learn about innovative concepts such as reinforcement learning, curriculum learning, and the integration of curiosity in AI systems. Gain a deeper understanding of ML training environments, cooperative rewards, and the limits of standard reinforcement learning. Investigate the potential of Long Short-Term Memory (LSTM) and the balance between extrinsic and intrinsic rewards in AI development. Uncover the challenges and opportunities on the path to creating more advanced, generalized artificial intelligence systems.
On the Road to Artificial General Intelligence
Overview
Syllabus
Intro
About Unity Technologies
What is Artificial Intelligence?
Intelligence in Biological Systems
ML Training Environments
Nature's Learning Method: Reinforcement
Chicken Crossing the Road: Tabula Rasa
2 Agents 1 Brain - Cooperative Rewards
Curriculum Learning: Progress
Enabling Long Short-Term Memory (LSTM)
Extrinsic and intrinsic Rewards
Limits to Standard Reinforcement Learning
The Quest for Surprisal: Curiosity (in Math)
The Good Puppy, Bad Puppy Method
ML-Agents Training Environments
The Road to Artificial General Intelligence
Taught by
GOTO Conferences
