newAIParadigms

newAIParadigms

Neuroscientist: The bottleneck to AGI isn’t the architecture. It’s the reward functions: a small set of innate drives that evolution wired to learned features of our world model, and that gives rise to generalization.

Neuroscientist: The bottleneck to AGI isn’t the architecture. It’s the reward functions: a small set of innate drives that evolution wired to learned features of our world model, and that gives rise to generalization.

The Hope architecture: Google's 1st serious attempt at solving continual learning

The Hope architecture: Google's 1st serious attempt at solving continual learning

A quick overview of the remaining research challenges on the path to AGI

A quick overview of the remaining research challenges on the path to AGI

The Continuous Thought Machine: A brilliant example of how biology can still inspire AI

The Continuous Thought Machine: A brilliant example of how biology can still inspire AI

[Analysis] Introducing Supersensing as a promising path to human-level vision

[Analysis] Introducing Supersensing as a promising path to human-level vision

Transformer Co-Inventor:

Transformer Co-Inventor: "There are already architectures that have been shown in the research to work better than Transformers. But to replace such an established architecture, being better is not enough. They need to be obviously crushingly better"

Ilya on the mysterious role of emotions and high-level desires in steering the brain's learning

Ilya on the mysterious role of emotions and high-level desires in steering the brain's learning

[Animation] In-depth explanation of how Energy-Based Transformers work!

[Animation] In-depth explanation of how Energy-Based Transformers work!

Probabilistic AI chip claims 10,000x efficiency boost. Quantum-style revolution (with real results this time) or just hype?

Probabilistic AI chip claims 10,000x efficiency boost. Quantum-style revolution (with real results this time) or just hype?

Father of RL and Dwarkesh discuss what is still missing for AGI. What do babies tell us?

Father of RL and Dwarkesh discuss what is still missing for AGI. What do babies tell us?

[Analysis] Deep dive into Chollet’s plan for AGI

[Analysis] Deep dive into Chollet’s plan for AGI

Crazy how the same thought process can lead to totally different conclusions

Crazy how the same thought process can lead to totally different conclusions

Fascinating debate between deep learning and symbolic AI proponents: LeCun vs Kahneman

Fascinating debate between deep learning and symbolic AI proponents: LeCun vs Kahneman

The PSI World Model, explained by its creators

The PSI World Model, explained by its creators

Casual discussion about how Continuous Thought Machines draw modest inspiration from biology

Casual discussion about how Continuous Thought Machines draw modest inspiration from biology

Browse Other Categories