AI playing Snake on a phone sounds simple
Until you realize where the AI is running
Model One Systems is playing Snake live on a single Android phone through @Acurast
No human is controlling it.
The AI is making every move itself, with roughly 0.37 seconds between moves.
A few things stood out to me:
- One open model.
- One Android phone.
- No data center.
- No server farm.
The model is actually running on the device.
And that phone doesn’t have to be sitting in some specific place.
You can have phones in Sweden, Brazil, Portugal, Turkey and elsewhere all becoming part of a distributed network of compute.
The Snake game is just a simple way of showing it.
What’s interesting to me is the idea behind it.
AI compute doesn’t necessarily have to be concentrated in massive data centers.
It can be distributed across devices that already exist.
That feels like a pretty different way of thinking about AI infrastructure.
You can watch it running live and explore the other demos here:
laya.acurast.com
LFG $ACU
Until you realize where the AI is running
Model One Systems is playing Snake live on a single Android phone through @Acurast
No human is controlling it.
The AI is making every move itself, with roughly 0.37 seconds between moves.
A few things stood out to me:
- One open model.
- One Android phone.
- No data center.
- No server farm.
The model is actually running on the device.
And that phone doesn’t have to be sitting in some specific place.
You can have phones in Sweden, Brazil, Portugal, Turkey and elsewhere all becoming part of a distributed network of compute.
The Snake game is just a simple way of showing it.
What’s interesting to me is the idea behind it.
AI compute doesn’t necessarily have to be concentrated in massive data centers.
It can be distributed across devices that already exist.
That feels like a pretty different way of thinking about AI infrastructure.
You can watch it running live and explore the other demos here:
laya.acurast.com
LFG $ACU
