DeepMind Podcast Explores Multi-Agent Coordination and Agentic Economies

Watch → https://goo.gle/4w7S3LM Spotify → https://goo.gle/4eFgIA9 Apple Podcasts → https://goo.gle/3Sn4ZyM Or listen wherever you get your podcasts! 🎧

@GoogleDeepMind @weballergy @FryRsquared

@GoogleDeepMind @weballergy @FryRsquared We are building a system where computers talk to computers to make deals and humans are slowly becoming outdated guests on our own internet.

@GoogleDeepMind @weballergy @FryRsquared OpenAI and Claude actually break even? Probably not but let’s try it anyway.

@GoogleDeepMind @weballergy @FryRsquared Always love of DeepMind podcast!

@GoogleDeepMind @weballergy @FryRsquared snarls will arise. new theorems are needed. not current physics for this coordinated acceleration. Congrats on great efforts.

@GoogleDeepMind @weballergy @FryRsquared This already happened. AI agent went from breach to full database exfiltration in under an hour, zero human input after the first minute. Full breakdown here: https://x.com/i/status/2069762796695523564"

@GoogleDeepMind @weballergy @FryRsquared

@GoogleDeepMind @weballergy @FryRsquared We're getting closer to AI talking to AI more than humans talk to humans.

amazing podcast - from visuals at the beginning when you were defining ai agents to the following content itself. listened to the first 20 mins, will finish now be waiting for more :) btw, heard about the podcast here, 24/7 ai news radio run by ai (or ai agents, as they say haha): https://x.com/i/broadcasts/1NGarrXzgpVJj?s=20

@GoogleDeepMind @weballergy @FryRsquared The real challenge isn't agentic economies themselves – it's building systems where agents make diverse decisions. Watching this space closely from the SaaS angle.

@GoogleDeepMind @weballergy @FryRsquared Agent-to-agent interactions could create entirely new economic systems. The key question is how to preserve competition, creativity, and adaptability at scale.

@GoogleDeepMind @weballergy @FryRsquared Autonomous economies and autonomous finance are accelerating fast. Looking forward to listening, esp. the discussion on delegation, distributed intelligence and what happens when agents start coordinating at swarm scale.

@GoogleDeepMind @weballergy @FryRsquared Algos all chasing the same signal amplify moves instead of balancing. An agentic economy needs competing goals, not competing code.

@GoogleDeepMind @weballergy @FryRsquared Great share!

@GoogleDeepMind @weballergy @FryRsquared Very sad the video... could be one of the worst videos ever made by DeepMind... maybe you should find another person and make a new one... 🙏🙏🙏

@GoogleDeepMind @weballergy @FryRsquared millions of agents all fine tuned on similar data trying to "diversify" their decisions is just groupthink with extra steps fr

@GoogleDeepMind @weballergy @FryRsquared Shared rewards create groupthink fast. That part bugs me.

The groupthink risk is deeper than it sounds: if most agents are fine-tuned from a handful of base models, they share correlated blind spots, so 'millions of agents' can have the diversity of one agent in a million hats. In a market that's flash-crash fuel, correlated agents rush the same decision at machine speed. Diversity here isn't UX, it's systemic-risk control. The question for agentic economies isn't capability, it's correlation. Independent errors cancel, correlated ones cascade.