I’m finding AI claims more plausible over time. This is partly because I’ve learned more, but partly because I think SIAI/MIRI has gotten better at communicating. In the past I thought the language was silly and anthropomorphic, with terms like “friendly” and “unfriendly”, so the scenarios didn’t strike me as plausible. Even if AI writers were trying to convey reasonable ideas, the language was offputting for a superficial reader. I hope MIRI will continue to use clear, objective language and cut back on jargon and jokes.
I’m finding AI claims more plausible over time. This is partly because I’ve learned more, but partly because I think SIAI/MIRI has gotten better at communicating. In the past I thought the language was silly and anthropomorphic, with terms like “friendly” and “unfriendly”, so the scenarios didn’t strike me as plausible. Even if AI writers were trying to convey reasonable ideas, the language was offputting for a superficial reader. I hope MIRI will continue to use clear, objective language and cut back on jargon and jokes.