Comment Re:Teaching AI to fail (Score 1) 49
Knowing how not to do a thing is also valuable. But just in general there were lots of emails, Strtegy documents, etc that would be valuable as how to make this thing even if the data inside is bad or the decision made on it was bad. Plus the customer info is valuable. Spirit tried to maximize extra purchases while keeping low initial costs. It failed eventually but the internal metrics at what was working and what was not would be very useful. They didn't immediately fail.