| |||
 
| |||
|
0188 - Instincts and Interactions - 2021.05.10 |
||
|
| |||
|
Comment: People with a firm cosmology and rigid moral framework (particularly, in my experience, monotheists) will often insist that other people must, likewise, have some sort of rigid moral framework, in order to live morally. To lack such a consistent structure but still insist upon right and wrong is, they assert, hypocrisy; those who do not have a specific system of rules to dictate and justify their behaviour may as well steal and rape and murder their way across the countryside, as there is nothing to stop them. Such an assertion, I will admit, felt correct - self-evident, even - when I was in my teens and early twenties. As I've spent time with the human experience, however, I've learned that our cognition is a teetering scaffold, built upon the back of a grunting animal. Morality may, in a textbook, be derived from the word of God or the rule of law, but, in practical effect, it is based upon either sympathy for the weak, self-sacrifice for the community, or fear of reprisal. All rules and laws are made retroactively to codify and justify this instinctual behaviour - and, crucially, to justify the aforementioned reprisal for those who stray from it. There are, therefore, two types of evil. The first is your common, pedestrian evil; the actions of someone who possesses insufficient sympathy or altruism or fear. Someone who doesn't care for the weak is cruel and callous. Someone who doesn't care about the community is greedy and destructive. Someone who doesn't fear reprisal is negligent and dangerous. Such people are mere criminals, looters and hooligans, mammals whose instincts are misaligned, who simply need to be either trained into proper behaviour or separated from society. The second type of evil, though, is far more insidious and deadly. The second type of evil comes from those whose system of morality - that elaborate teetering scaffold - is cantilevered so far out from the mammal that it no longer connects. These are your inquisitors, your slave-masters, your hellfire preachers and fanatical generals. These are the grey-faced logicians who insist that their history of facts and chain of logic are inviolable. These are the kings and popes and bosses who counsel their followers to suppress their natural love and fear, for these are the mere instincts that may sway you from doing what their logic insists, with ironclad certainty, is Right. Such monsters cannot be rehabilitated - they can only be defeated and plowed under by the march of history. The problem here is that an AI is, by definition, all scaffold, no beast. In much the same way that driverless cars (at the time of writing) still require a human being with instincts and "common sense" in the driver's seat, AIs may require a mammal to sign off on their ethical decision-making for the foreseeable future. Of course, as we've seen, a human sitting behind the wheel of a car that makes 99.95% of its decisions correctly will very quickly drift into complacency. All of this is to say, of course, that a counselling program that can help you reconcile with a spouse or train a dog will be of little use when you are attempting to debug the user interface of an android. Instinct is important. |
||
|
Transcript: --------------------------------------------------------------- |
||