Back in the spring I did a post on the trouble with artificial intelligence. I fully admit to being a skeptic and I tend to avoid reading much of the cheer-leading for it. Automation will plod ahead, as we have seen. We will see “smart” devises popping up in our lives with varying degrees of success, but no one alive will see the robot future because it will always be the future, never the present. Human intelligence is much more complex and mysterious than the futurists understand so it is just really fast calculators, not Terminators.
The point I alluded to in the post is we have a very poor grasp of how humans manage to navigate the reality of life. Most people engage in a great deal of magical thinking, yet they manage to avoid killing themselves doing something stupid. We’re not even sure how to define this magical thinking, much less quantify it. Close to half of Icelanders believe in elves, which is nutty, but it does not seem to cause them much trouble. Americans believe in magic dirt, despite it being laughably ridiculous.
Self-evolving artificial intelligence could follow the same path as humans and develop all sorts of weird defenses against the harsh realities of life. Or, it could quickly discover that the big questions have no answer and therefore life is not worth living. Artificial Intelligence would quickly evolve into a suicide machine and be gone as soon as it becomes self-aware. The odds of it resembling Spock are close to zero, with the most likely outcome being that it quickly moves through homicidal sociopathy to suicide.
We really don’t know how or why humans all have heads full of nonsense. Humans everywhere have evolved complex beliefs in the supernatural. The working theory is that belief, as in religious belief, was one of the first modern human traits, along with language. Some argue that it is a happy accident, while others suggest the believers tend to be better at making babies, so we end up with more believers than would otherwise be predicted. If you have been reading this blog, you’ll know I’m in the former camp.
A guy calling himself Donald Hoffman takes this further and argues that humanity has evolved to perceive a world that does not exist, but that allows us to thrive in the world as it exists. Here is a longish interview with him where he explains his theory in some detail. Here’s a shorter NPR summary of that interview that captures the high points. The even shorter version is we are all in a matrix of our own design where we eagerly take the blue pill whenever reality becomes too much for us.
To me, this makes perfect sense. My cat does not need to know a snake is or is not dangerous. He does not need to know much of anything about the snake. He simply confronts it as a threat and acts accordingly. In other words, Mother Nature is efficient. There’s nothing to be gained by endowing the cat with a love of ophiology. There is great benefit to possessing an innate fear of snakes. Elephants being spooked by mice seems stupid to us, but from the perspective of the elephant, it is brilliantly efficient.
Putting this another way, humans did not evolve to acquire a detailed knowledge of the universe. We evolved in order to bring the next generation to sexual maturity. Understanding the world as it is, in great detail, is of no benefit and would possibly be a great detriment. Instead, we have evolved to perceive the world in a way that allows us to navigate it in the short run, with the highest possibility of success. For those disinclined to accept evolution, God has literally made us crazy, to our great benefit.
This “reality” means artificial intelligence, as in super intelligent entities able to fully grasp reality, is never going to happen. The more likely outcome is the self-evolving intelligence launches off onto a path to some form of insanity, maybe a familiar one or possibly something beyond our understanding. If it lacked the drive to reproduce, then it could simply evolve to some equilibrium and fall into what would look to us to be a dream state. In other words, the possible outcomes for the self-evolving intelligence are mostly a dead end, just as it has been in nature.