Episode Details

Back to Episodes
What Comes After AI Transformers? (Ep. 531)

What Comes After AI Transformers? (Ep. 531)

Episode 531 Published 4ย months, 4ย weeks ago
Description

The discussion sets the stage for exploring what comes after transformers.


Key Points Discussed


Transformers show limits in reasoning, instruction following, and real-world grounding.


The AI field is moving from scaling to exploring new architectures.


Smarter transformers can be enhanced with test-time compute, neurosymbolic logic, and mixture-of-experts.


Revolutionary alternatives like Mamba, Retinette, and world models introduce different approaches.


Emerging ideas such as spiking neural networks, Kolmogorov Arnold networks, and temporal graph networks may reduce energy costs and improve reasoning.


Neurosymbolic hybrids are highlighted as a promising path for logical reasoning.


The challenge of commercializing research and balancing innovation with environmental costs.


Hybrid futures likely combine multiple architectures into a layered system for AGI.


The concept of swarm intelligence and agent collaboration as another route toward advanced AI.


Timestamps & Topics


00:00:00 ๐Ÿ’ก Introduction and GPT 5 disappointment


00:02:00 ๐Ÿ” The shift from scaling to new paradigms


00:04:00 โš™๏ธ Smarter transformers and test-time compute


00:05:20 ๐Ÿš€ Revolutionary alternatives including Mamba and Retinette


00:06:20 ๐ŸŒ World models and embodied AI


00:06:58 ๐Ÿง  Spiking neural networks and novel approaches


00:11:00 โ›ต Exploration analogies and transformer context challenges


00:12:20 ๐ŸŽฎ Applications of world models in 3D spaces and XR


00:16:45 ๐Ÿ”— Neurosymbolic hybrids for reasoning


00:19:00 โšก Energy efficiency and productization challenges


00:24:00 ๐ŸŒฑ Balancing research speed with environmental costs


00:31:00 ๐Ÿ“‰ Four structural limits of transformers


00:35:00 ๐Ÿ“š RKV and new memory-efficient mechanisms


00:37:00 ๐Ÿ“ Analogies for architectures: note taker, stenographer, librarian, consultant


00:41:00 ๐Ÿ•ต๏ธ Transformer reasoning illusions and dangers


00:44:00 ๐Ÿ”ฌ Outlier experiments: physical neural nets, temporal graph networks, recurrent GANs


00:49:00 ๐Ÿงฉ Hybrid architecture visions for AGI


00:53:30 ๐Ÿ Swarm agents and collaborative intelligence


00:55:00 ๐Ÿ“ข Closing announcements and upcoming shows


The Daily AI Show Co-Hosts: Andy Halliday, Beth Lyons, Brian Maucere, Eran Malloch, Jyunmi Hatcher, and Karl Yeh

Listen Now

Love PodBriefly?

If you like Podbriefly.com, please consider donating to support the ongoing development.

Support Us