Skip navigation
Help

Global Brain

How to create a global brain in only a few lines of code

This is where my research was going years ago and I found so many interesting things that I forgot that's why I was doing it. Here's the summary of the plan:

User Interface can be anything which is statistically balanced and has continuous input and output of at least 1 dimension between each person and their computer. A simple example is the speed they are moving the mouse, if its increasing or decreasing at the time, and an output could be some music which is playing becomes a little slower or faster at the time. It could be more complex things like realtime video, evolved audio, Nintendo Wii controllers, Kinect, Emotiv Epoc or OpenEEG mind reading game controllers, or many other things. The User Interface is a stream of vectors in and vectors out, of at least 1 dimension, through any devices. If there is any audio or video, that is part of the User Interface. The core idea is a kind of math and is calculated independently of any game content which players create while in the game.

N people play the game at once, streaming data to eachother's computer through the Internet as it was all 1 system with many inputs and outputs as paths of information flow between the players.

The output to each player is a prediction of the next input of that player. The player must hear/see/experience the output in some way so it affects their state of mind.

The combined inputs of all players are used to predict the combined outputs of all players. This can be done many ways. A bayesian network should work well for this since it calculates using the math of conditional-probability and scales up efficiently.

Here's what makes it work extremely more than the intelligence of the AI or any 1 player:
Since the bayesian network calculates relevance of inputs and outputs to its prediction accuracy, whichever inputs of other people are most useful (combined in some statistical way) to predict the next few inputs of this local person, will gradually be given more influence here, and because of that this local person, who "must hear/see/experience the output in some way", will tend to become more statistically relevant for the AI to use their inputs to predict the other peoples' next few inputs, and the feedback loop is complete and amplifies peoples' ability to play the game in a way that helps the AI use people to predict other people.

In this feedback loop of N people, without needing conscious knowledge or intent of it, people will unavoidably be influenced toward flowing their thoughts together because the set of all possibilities where that does not happen is partially cancelled-out by the bayesian network.

Depnding on the accuracy of whatever kind of AI does these predictions and is the "glue code" for networking our minds together, and how skilled people become at the game, a superintelligence is somewhere along this research path and it will be made of the minds of billions of people and computers flowing thoughts together at the subconscious psychology level.

This is the simplest way to build a superintelligence. My research years ago took a different direction in finding User Interfaces, like Audivolv, BayesianCortex, and Physicsmata (all open source), and now I have a good idea of how to put it all together. We can proceed with these experiments toward thinking more like a global brain.

Does anyone have idea on what kind of game it should be? The research path leaves many possibilities.