The AI Browser
Building the Foundation of Human Browser Behaviour
The goal of the AI Browser is to train an AI to do all that a human being would do. Not to create a goal, but to teach it purely the foundations. And then let it play with AI itself. Within guardrails.
What it's about
This is about exploring how much can we bring an AI to interact with the web like a human being when it knows the basics. When I was young, I would teach math classes to those that struggled with math. They learned the basics and I as I knew the basics that good afterwards, I was able to solve the advanced topics too even though I never learned them actively.
That is the idea with the AI Browser. If the AI knows to click a button, open a sidebar and hold a mouse a bit longer for a Captcha, how well is it able to navigate through the web itself. And then, if it manages these actions, what implications does it have? For the security of the web? And the way of how we interact on the web?
The Vision
The why is similar to the what it is about - learning what is feasible and how fast and well is it possible to then after that foundation let the AI do specific actions based on what it is trained on. Potentially connecting that concept and product with the AI Agentic Game Play and constructing and viewing the web as a 3D world/a world for an agent to play within.
Even further thought - is it possible to let an AI step out of one world and into another? How would that look like, why would it be possible and why now?
Best For
Researchers, curious AI cats, innovators, boundary pushers - all those wanting to experiment and research on what is possible with AI.
Tags
