Subvolts take AI might evolve like life, discovering science itself, instead of just getting bigger. Feb 24, 2026 - Developers
- Learn about alternative AI development strategies like model merging and multi-agent systems beyond scaling.
- Business
- Understand how AI might evolve autonomously, potentially leading to new scientific discoveries and capabilities.
This interview explores a novel approach to AI development, focusing on evolution and collective intelligence as potential breakthroughs.
#323 David Ha: Why Model Merging Could Be the Next AI Breakthrough
Audio streams from Craig S. Smith. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
David Ha, CEO of Sakana AI, discusses AI's next breakthrough potentially coming from evolutionary strategies and collective intelligence, not just scaling models. The "Eye on AI" episode explores model merging, multi-agent systems, and AI's potential for open-ended discovery beyond human knowledge. Ha suggests AI could evolve and discover science autonomously.
Instead of making AI models bigger, this episode explores letting AI evolve like living things. This could lead to AI discovering new ideas and science on its own, potentially changing how we build intelligent systems.
This episode offers a look at alternative AI development paths beyond simply building larger models. It's for researchers and AI enthusiasts interested in emergent intelligence and automated scientific discovery. The concepts discussed, like multi-agent systems and evolutionary AI, offer a different perspective on the frontier compared to the massive model scaling often seen at labs like Google Gemini and OpenAI.
- HostCraig S. Smith
- GuestDavid HaCo-Founder and CEO, Sakana AI
- Guest: David Ha, Co-Founder and CEO of Sakana AI
- Episode Title: Why Model Merging Could Be the Next AI Breakthrough
- Show: Eye on AI
- Length: 57 minutes
This episode is sponsored by tastytrade. Trade stocks, options, futures, and crypto in one platform with low commissions and zero commission on stocks and crypto. Built for traders who think in probabilities, tastytrade offers advanced analytics, risk tools, and an AI-powered Search feature. Learn more at https://tastytrade.com/ Artificial intelligence is reaching a turning point. Instead of building bigger and bigger models, what if the real breakthrough comes from letting AI evolve?
In this episode of Eye on AI, David Ha, Co-Founder and CEO of Sakana AI, explains why evolutionary strategies and collective intelligence could reshape the future of machine learning. We explore model merging, multi-agent systems, Monte Carlo tree search, and the AI Scientist framework designed to generate and evaluate new research ideas. The conversation dives into open-ended discovery, quality and diversity in AI systems, world models, and whether artificial intelligence can push beyond the boundaries of human knowledge.
If you're interested in AGI, evolutionary AI, frontier models, AI research automation, or how AI could start discovering science on its own, this episode offers a clear look at where the field may be heading next.
Stay Updated: Craig Smith on X: https://x.com/craigss Eye on A.I. on X: https://x.com/EyeOn_AI (00:00) AI Should Evolve, Not Just Scale
(03:54) David's Journey From Finance to Evolutionary AI
(10:18) Why Gradient Descent Gets Stuck
(18:12) Model Merging and Collective Intelligence
(28:18) Combining Closed Frontier Models
(32:56) Inside the AI Scientist Experiment
(38:11) Parent Selection, Diversity and Innovation
(49:25) Can AI Discover Truly New Knowledge?
(53:05) Why Continual Learning Matter
- Model Merging
- David Ha
- Combining different AI models to create a new, potentially more capable one.
- Multi-agent systems
- David Ha
- Multiple AI agents working together to solve problems.
- Monte Carlo tree search
- David Ha
- A method for making decisions in AI, often used in games, by exploring many possibilities.
- AI Scientist framework
- David Ha
- A system designed to help AI generate and test new scientific research ideas.
55 min42 min51 min34 minUkraine's Secret Weapon: The Points System Winning the Drone War | Andrii Hrytseniuk
WithAndrii Hrytseniuk
1 h 05
1 h 04
Questions people ask
- What is the main idea for AI's next breakthrough?
- The breakthrough may come from letting AI evolve using evolutionary strategies and collective intelligence, rather than just building bigger models.
- What concepts are explored in the episode?
- The episode covers model merging, multi-agent systems, Monte Carlo tree search, and the AI Scientist framework for generating research ideas.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Craig S. Smith.