DeepMind's wet dream is M3-Agent's reality: how long-term multimodal memory is modelling the real world
AI Today17 Elo 2025

DeepMind's wet dream is M3-Agent's reality: how long-term multimodal memory is modelling the real world

Google DeepMind's Demis Hassabis and his team have a bold mission: penetrating the 4D chess game that's AI embracing our ever-changing biological, physical world.

Taking a snapshot is one thing. Remembering the molecular topology and their constant changes of state is truly what separates fact from fiction.

It seemed like an impossible target to hit. Until M3-Agent, the work of researchers associated with ByteDance at Shanghai Jiao Tong University, showed up with long-term multimodal memory - allowing the agent to see, hear, remember, and reason just like humans.

M3-Agent's potential is groundbreaking.

Here are just three use cases that will blow all our minds:

  • Autonomous robotics: Robots in homes or warehouses remember object locations, user habits, and past errors, adapting tasks dynamically, such as a caregiver bot recalling a patient's routines for personalized aid
  • Enhanced surveillance: Security systems analyse live video/audio feeds, building memory of normal patterns to detect anomalies, predict threats, and reason through scenarios, like identifying intruders based on historical behaviours
  • Personalised education: AI tutors process student interaction videos, remember progress and misconceptions over time, and deliver tailored lessons, such as adapting math explanations from weeks of observed struggles.


Read the paper: Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory.

Jaksot(90)

Prompt engineering masterclass

Prompt engineering masterclass

Here at AI Today, we know how to listen. We spent hours analysing Lenny Rachitsky - host of Lenny's Podcast - interviewing pro prompt engineer Mike Taylor to bring you this deep dive into all the tech...

30 Loka 202416min

How to make a ton of Monet with AI art...

How to make a ton of Monet with AI art...

Botto's 15,000 curators are celebrating a big win this week after six of their carefully-chosen, pixel-pushed masterpieces, sold for more than $350,000 at a Sotheby's auction in New York. It's a story...

29 Loka 202415min

Autonomous agents: Rebooting your business the right way

Autonomous agents: Rebooting your business the right way

Imagine if you had massive balls - crystal ones - to accurately forecast future business needs. That's one of the thousands of ways autonomous agents - popularised in organisations of all sizes throug...

27 Loka 202413min

Lose your RAG

Lose your RAG

Retrieval augmented generation is how we used to chunk content in huge corpuses of data. Now there's a new sheriff in town - contextual retrieval preprocessing, or contextual RAG. No more relying on k...

25 Loka 202416min

#FutureOfWork: AI as the enterprise nervous system with Microsoft's new Copilot

#FutureOfWork: AI as the enterprise nervous system with Microsoft's new Copilot

Let's take a look at how the latest version of Copilot can change the game for your business. Imagine a manufacturing company developing a new electric vehicle (EV) charging station. This complex proc...

24 Loka 202418min

How to build a video editor with Anthropic's Claude AI

How to build a video editor with Anthropic's Claude AI

If there's anyone left in the world yet to be convinced AI is changing it, have a chat with Meng To (@mengto on X). He just wrapped up Dreamcut.ai - what he calls his perfect video editor - after spen...

23 Loka 20249min

Claude Takes Control: AI That Uses Computers Like We Do

Claude Takes Control: AI That Uses Computers Like We Do

Forget everything you thought you knew about AI assistants. We're not talking simple chatbots that can barely string a sentence together. Claude 3.5 Sonnet, the latest iteration of Anthropic's groundb...

22 Loka 20249min

Solver brings full self driving to AI coding

Solver brings full self driving to AI coding

Engineering teams are frazzled. And we've all been down the Cursor, Aider, Cline, Bolt, and Replit rabbit roles questing for AI coding nirvana. But there are more potholes in the process than a worn-o...

22 Loka 20249min