Comments by "Adam Donovan" (@adamdonovan4071) on "Contra Scott Alexander on AI Safety Arguments" video.

  1. I suggest that another lens for the distinction between the true and untrue panics is the true ones (medical and geopolitical) relate to external actors, whereas the untrue ones are concerns about internal actors (what will we do if the spinning wheel takes our jobs for example)…AI, specifically general AI is inventing a new actor; it is absolutely an external actor to human society. There is definitely no way to anticipate how a general AI with adequate power to intentionally harm people will assess humans and whether or not they are “worthy” of survival. My concerns with AI are not regarding job losses, they are regarding survival. Maybe an AI will find us “useful”, but maybe it will not. In this sense, that we are introducing a vector heretofore unexposed to humanity; the AI would be better assessed as a virus, pathogen, or geopolitical actor (capable of initiating its own will) rather than as a “technology”….if we want to treat a specific AI in use for highly rote specialized tasks as a technology that seems right; but at the point where it is making value assessment the watershed moment has passed and it is now an actor of will, not a technology.
    3
  2. I don’t disagree that watershed moments, by their very definition, break historical norms; however I will also point out that we didn’t destroy ourselves, so it could be argued that they were in fact correct…something becoming possible does not make it inevitable. New hardware does not change the software; but does usually mandate a software update. The difference with AI, is that humans will no longer be the actor, and it will not necessarily put the same value on human life that the folks in control of the nukes during the MAD era had.
    3
  3.  @danielr5343  you can call it freak coincidence, but I suggest it is actually a human desire to survive. AI will never care if we survive, at least no programmer will ever be able to ensure that an AI which can adapt itself will maintain a desire for humans to exist. It takes a very “special” human to desire for humans to not exist, it takes a very “special” AI to desire for humans to exist. The generators are gambling with every human life, and when they lose, we all pay, but until they use, they get paid. It is an extremely bad path to choose, worse path for others to forces everyone else down.
    2