Actually, the BCS was entirely human-created and human-driven.
Three main components: the polls, the computer ratings, and strength of schedule.
-- Both polls (AP and Coaches) were determined entirely by human voters.
-- The computer component was driven by computer algorithms created by humans, like Jeff Sagarin, Jeff Anderson, and Chris Hester. Humans made all the decisions: they said to the computers, "take these updated numbers I give you, multiply them by that, and then add this, and tell me the result." They were simply programmed calculators. Had nothing to do with AI or deep learning. It was the humans making them who created the criteria and weighed the relative importance of this and that.
-- Finally, the strength of schedule component was the most pre-programmed bit of all, with the least human variability, but it was still set by a human-decided formula. Add up the win-loss record of each team's opponents as well as the opponents' opponents. Then rank order the results by the team who played teams with the best records.
But all of that is still human input, human-selected criteria. The difference between it and a committee like the current CFP was simply that the rules and criteria were set in advance with no allowance for judgment calls at the last minute.
