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)

Room for agentic AI? How hotels become smooth operators with the technological touch

Room for agentic AI? How hotels become smooth operators with the technological touch

AI Today creator Dave Thackeray today presented his own deep dive into how agentic AI is ready to be the key to efficient hotel operations - giving staff more time to deliver exceptional guest experie...

3 Kesä 202543min

Safe or just plain woke: Anthropic's Claude 4 system card

Safe or just plain woke: Anthropic's Claude 4 system card

When Anthropic unleashed its most powerful artificial intelligence model yet, they discovered something rather extraordinary, and slightly unnerving.Claude 4 Opus developed an unexpected habit of tryi...

3 Kesä 202519min

Mary Meeker's AI Trends

Mary Meeker's AI Trends

Hugely important work. But what does it mean to us? Today our hosts created their own company imagining how insights from this celebrated report would apply to the modern business environment.

1 Kesä 202520min

AI to HR: Welcome, intelligence optimisation!

AI to HR: Welcome, intelligence optimisation!

What happens to the People team when it's juggling bodies AND bots?Thanks for listening to this special episode of AI Today. Read along with the show, here.

25 Touko 202510min

25 ways to put AI agents to work - right now!

25 ways to put AI agents to work - right now!

We've been waiting a hot minute for some genuinely useful AI agent case studies to drop.Now we have 25 on our plate.Take a listen to the highlights reel and then download them for yourself:https://www...

21 Touko 202513min

Google I/O 2025: What happens now?

Google I/O 2025: What happens now?

Read the full story here:https://medium.com/@DaveThackeray/a-world-beyond-google-i-o-2025-ea56bcd5e208We're on the cusp of some major announcements that will send shockwaves, and a spike in defibrilla...

19 Touko 202516min

Hallucination solution : Customer service ready for revolution!

Hallucination solution : Customer service ready for revolution!

Researchers have made huge strides fixing bad trips for AI.One of the latest breakthroughs is attentive reasoning queries (ARQs).You can see them in action using the open source Parlant application.Wh...

15 Touko 202518min

Hallucination: a bitter pill to swallow

Hallucination: a bitter pill to swallow

AI hallucinates 100% of the time. That's by design - without hallucinating the next word, this transformer architecture wouldn't exist.Thankfully, LLMs built for general purpose applications are right...

13 Touko 202530min