Main Page Contact Register Log In

I agree. I'd like to examine those underlying processes to see what would allow an AI to turn evil quickly.

In the VA Hospital case - one mistake was that the VA was also tasked with measuring itself against its objectives (tied to employee performance awards). This seems like a common problem with many US government programs - take the EPA which both enforces and measures.

Added to this, as Deming pointed out a long time ago, they typically use the same information systems to measure their performance as they use to manage the process.

The result is that the data becomes hopelessly corrupted by the need to look good.
Replies (0)