Some time ago I heard a story about AI research. The researchers had set up a neural net and busily spent their days entering “facts” into the computer. Each night, the computer would chew on these “facts” and spit out what it figured out, essentially spitting out its interpretation of how, and what, the world was.
One morning, it declared, “All people are famous.”
To the researchers, this was a puzzle — until they realized that they’d begun entering information about people into the system and that, thus far, they’d only chosen and entered “notable” individuals.
To the computer/AI, it made clear, perfect, logical sense. It only knew of famous people. Thus, everyone must be famous.
While I don’t think it was their intention, the researchers built a pretty good example of how our own brains work.
Though sometimes we are admonished to “read between the lines,” our brains are always doing just that. They take all the vast amounts of information that comes in, parses it, organizes it, and looks for patterns… and then goes even further beyond to look for logical truths. “If such is such, and such is also such, then it follows that…”
To once again quote the great Carl Sagan: “The brain does much more than just recollect, it inter-compares, it synthesizes, analyzes, it generates abstractions.”
Abstractions, deductions, and truth/realities that totally fit with whatever knowledge and experience it has at that point in time.
This is all great, except that we don’t know our brains have done that. And that from thereon out, our brains will filter our new experiences and observations through that truth it already knows, even hiding things from our consciousness. And even more so that we will take many actions based on all those, quite potentially flawed, deductions. Sometimes it will work out. Sometimes our actions will be downright unproductive.
Thanks to that triple whammy, it can be tough for our patterns and predictions to get updated with new knowledge and experiences that, should at least, be coming in all the time. If we’re lucky, a different logical deduction will emerge and compete with an old one such that they balance each other out. Or we may get a half-update, where the brain still partially holds onto the vestige view, ready to jump back to it at the earliest “confirmation.”
In moments of our most desperate want, deductions can collide to create twisted logics of epic proportions, with epic(ally poor, often) results.
But by stepping back and choosing to go into a series of inquiries to do some heavy re-examination, we give our brains a chance to go back to the primordial and recalculate. By taking ownership of our views and deductions and realities we gain agency to revise them. We can come up to date with our stories so that they are in line with where we are today and where we want to go, crafting them so that they serve us well.
Then we can show those AIs how its really supposed to be done…