Imagine an AI that doesn't just answer, but thinks like a hacker. Now, Claude can do just that, but for good.
·1 min·5·Beginner
03Games provide interactive simulations and structured environments, vital for developing Artificial General Intelligence (AGI).
0101
What's wrong with LLMs for AGI?
Current artificial intelligences are absolute wizards with words. They write, summarize, and create incredible texts. But there's a tiny catch: they don't really grasp how things move in the real world. They don't know a cube rolls differently than a ball.
Models like ChatGPT and Claude excel at text generation but struggle with understanding the spatial dynamics of objects or basic physics laws. This is a huge gap if the goal is to create truly "general" intelligences, capable of adapting to any situation. What's the point of an AI that writes perfect essays but can't open a door?
0202
Why are video games the unexpected hero?
The solution to this lack of "physical common sense" might have been under our noses for decades: video games. 3D environments, simulated physics, complex interactions, and immediate consequences. They're a perfect training ground for an AI that needs to learn how to "live" in the real world without getting hurt.
📬 Enjoying this article?
Get the best AI news every week, straight to your inbox.
General Intuition is a startup exploring the use of video game data to overcome current limitations in artificial intelligence. Imagine an AI learning the basics of gravity in Minecraft or navigating a virtual maze, then applying those intuitions elsewhere. Pretty clever, right?
These virtual worlds offer a constant stream of structured, interactive data. They allow AI to experiment, fail, and retry millions of times without cost or risk. It's like sending a child to school, but with a perfect life simulation, without them breaking anything in the living room.
0303
A future with smarter, less "bookish" AI?
If General Intuition's idea works, future AI could have a much more solid physical intuition. It won't just process texts; it will truly understand how to interact with objects, predict trajectories, and move through space. Goodbye AI that answers everything but can't hold a coffee cup.
The ultimate goal of this approach is the development of Artificial General Intelligence (AGI), capable of learning and applying knowledge across various contexts. This doesn't mean AI will stop reading books. It means that, besides learning grammar, they'll also have a physical "common sense." Maybe that's the key to robots that don't trip you up.
NUOVO
