Imagine the chess-playing supercomputer Deep Blue. Now eliminate all of the chess algorithms in order to install your own.
No, scratch that: retain all of the chess playing algorithms, but fragment them throughout its hard drive. Now in between those fragments install the fragments of your own needlessly complicated algorithm. This algorithm will be capable of calculating 1+1. The system is already capable of doing this, indeed, it is necessary for the functioning of the individual components of your new algorithm. Nonetheless, all other calculations must use this algorithm in order to calculate 1+1 for any purpose.
Now, imagine you have a fleet of Deep Blue supercomputers, all arranged as before. Thousands of them. And they are scattered across the world, all networked together via dial-up modem. Some of them play chess with each other while calculating and communicating for the purposes listed. Actually, all of them are playing chess, and often more than one game.
This network is used to calculate and document the following:
- How many Deep Blues are in the Network. They're not allowed to count.
- How much power is drained in calculating 1.
- The projected telephone bill accumulated in calculating 1-10, using 1995 figures.
- Ways of improving the efficiency and usefulness of the network.
- Reasons not to employ any of the solutions proposed by 4.
- Ways of changing the network in such a way that will have either no effect on efficiency or drastically reduce efficiency.
- Reasons why it is absolutely essential to immediately employ all of the "solutions" proposed by 5, at once, in different regions of the network, without communicating individual changes to the rest of the network.
- How long it has taken to calculate all of the above.
- How long it has taken to calculate how long it has taken to calculate all of the above.
- How long it has taken to calculate how long it has taken to calculate how long it has taken to calculate all of the above.
As you can see, the last one introduces an infinite regress into the system that ensures that the Deep Blues will have employment forever.
Now, this network forms what is known in the AI research community as a "neural network", which mimics the neurology of the brain, and each node of the network has the processing power of one Deep Blue supercomputer. From this incredibly slow-functioning network a genuine intelligence emerges. Whether or not it is self-aware is unknown, certainly, even if it is aware of its existance it is definitely not
self aware in the common usage of the term.
This emergent artificial entity, composed of thousands of networked Deep Blue supercomputers, has the sum total intelligence of the average bee.
~
And that is how bureaucracy works.
