In the 60′s and 70′s there was lots of hope and high expectations (never good things, FYI). Artificial Intelligence (AI) was going to be conquered with relative ease and by the turn of the millennium we’d have self-aware machines helping out humanity wherever they could out of the kindness of their heart, or they’d have taken over society and enslaved us all. Either way, everyone was certain it was going to happen Any Day Now.

That didn’t happen. Autonomous robots are still a fanciful thought with the closest approximations of thinking and feeling robots merely mimicking emotions they’ve been taught to mimic and parsing the expressions on our faces through complex analysis which ultimately comes down to further training of what emotions a certain kind of face means.

There are some efforts out there that rely on emergent properties popping up in simple loops of code which are initially taught a few base commands some of which would allow the code to modify itself. These are slow going but I think they are the best bet right now.

A few years ago I was considering a minor in cognitive neuropsychology, primarily because I was interested in AI and I wanted to try it from a different angle. Rather than come at it from a mathematical deterministic manner, I began to think about AI development from an evolutionary standpoint.

Intelligence didn’t come from nowhere; it took thousands of generations of incremental improvements, both physical and mental, to get to the level of humans, or even the level of dogs. Which is why self-modifying programs seem like the best bet of the available options, but I think there is one thing that self-modifying programs lack that could be crucial in developing truly independent and self-aware intelligent machines: childhood.

I’m sure that my stance has either been attempted enough times to be proven useless or is actively being researched by people in the field who know much much more about all this stuff, but I think that the AI researchers out there need to start looking at creating families. Instill in the base code of the first generation of programs a need for procreation and a few other basic operations and let life flourish or crumble.

I’m not saying that this is easy, but I think it will be the method by which substantial artificially intelligent machines will be created. Rather than create something in our own image, we must generate an environment conducive to development and allow it to persist.