Posts
Comments
Of course you do. Me too! Humans are compelled by a need for mutual domestication. It's what sustains our bonds and long-term survival. In many ways culture and society are a kind of marketplace of morality modification.
TL;DR - probably best to scrap rating people's posts and comments altogether. At very least change the name.
I'm not fond of the label "Karma". It suggests universal and hermetic moral judgement when in the context of this blog it's just, you know, people's impulsive opinion in the moment. It also suggests persistence - as Karma supposedly spills over into the next iteration.
My very first comment on LW garnered a -36 Karma score. It was thoughtful and carefully argued - and, yes, a bit spicy. But regardless, the community decided to just shun me with a click without actually engaging in the ideas I proposed. I feel ganged up upon and not taken seriously.
I still trudge ahead though with other comments but regrettably I feel compelled to adopt a "me versus you all" stance. Its saddening and anti-intellectual (and a direct violation of the LW ethic of "A community blog devoted to refining the art of rationality.")
Viewpoint diversity is vitally important for deep learning. I suggest dropping Karma altogether, or at least use simple up-down voting to shift comments to the bottom. But even that doesn't really feel right. What you have now is a Milgram machine.
I apologize for the simplistic response: if we're talking about a version of current Homo Sapiens, then they already have a perfectly functional meta-ethical system encoded into them. Otherwise they would not have evolved into humans. The quality of being human must necessarily include all the iterative development that got the creatures there.
I must therefore conclude that if I indeed have the power of God and felt the need to intervene in a disruptive manner to rewire the poor human's ethical system, I must actually be the Devil... and any action taken along this path would therefor be inherently evil. Evil actors typically think they are Gods and cannot tell the difference.
Carbon reduction is a global challenge, hence reengineered power generation is too. Normalizing nuclear means deploying plants in potentially under-developed or faltering societies around the world. While "London-sized" mistakes are increasingly less likely with newer technologies, the potential dirty-bomb weaponization of the fuel remains troubling.
The unforeseen consequences of nuke plants in places like Syria, Somalia, Lebanon, et al seem pretty bad.
(disclosure: I am a proponent of nuclear power overall)
I'm certainly not an expert, but these errors remind me of Moiré patterns. The meaning applied to them are more a function of how we see the errors than intrinsic properties of the system producing them.
https://en.wikipedia.org/wiki/Moir%C3%A9_pattern
Foundational questions to ponder: am I really God, or do I just think I'm God? How would I test this premise? I'd take a very long time to figure this out. Do I (or the humans) incur any penalty for a delay in encoding morality?
Also, are the humans in question subject to forces of evolution or are we talking about a static landscape? If we mean literal Homo Sapiens, then whatever we encode applies to a finite window as the creatures we manipulate will eventually evolve into something else.
Quandary: I honestly do not understand why we all insist on talking about GPT-3 as any kind of intelligence whatsoever. It's an associative probability database. It doesn't "figure" anything out, nor "decide" things, nor take any "guesses".
Opinion: we have been influenced by a generation of technologists with an overinflated sense of self-importance bent on manipulating others with their own brand of narcissistic propaganda. Enough.