Singularity University – Day Two – Evening – Ray Kurzweil
Ray Kurzweil gave a lecture on the "Critics of the Singularity" Tuesday evening after dinner.
In the Singularity University’s usual style, these "lectures" have more of a "discussion" feel to them, as students are encouraged to ask questions and challenge any concepts being presented if they disagree, in order to create a more interactive learning environment.
"The whole point of the Turing test is that you really have to have an understanding of language in order to pass it," he explained. "But if a machine can pass the turing test, that’s not a demonstration that a machine is conscious."
"But how will we know?" The students asked.
Ray then explained why we probably won’t ever be able to know for sure, as there is no way to verify "consciousness," in humans, machines, or anything else.
Here is a partial transcript:
"When we get to that point, the whole point is that these machines will be convincing in their mastery of human emotional responses. But it is just a simulation of consciousness? How could you possibly measure that?"
I know when a person seems conscious – not all people seem conscious (laughter) – but, when a human seems conscious – that thee is a conscious person there, just like me. But "consciousness" is a synonym for "subjectivity." And "science" is a synonym for "objective observation and inductions from our observations," there’s a conceptual gap, in my view, between "subjectivity" and "objectivity," and I can’t imagine an objective test that really would fundamentally philisophically demonstrate the consciousness of another entity.
We have these disagreements today. We have kind of a shared assumption of human consciousness, but that assumption breaks down as we go outside of human experience. We don’t get to have arguments about the machines because there aren’t machines that appear to have that complexity of behavior. But people disagree about animals. Some people feel that animals are just operating by "instinct" (motions fingers as quotation marks), an old fashioned machine concept. You have this stimulus response in a simplistic way, that’s interacting, with essentially nobody home. Other people think that higher level animals are conscious. I think my cat is conscious. Other people don’t agree, but they haven’t met my cat. (laughter)
But the whole issue of animal rights boils down to "are these animals actually conscious? Are they actually suffering? And, ya know, to the extent that we feel like we see human-like reactions, there’s some consensus origin that they’re conscious. But that’s also a human-centric position, because we perceive everything with a human-like behavior.
There are some animals, like giant squids, who have a very complex, large brain, but their behavior’s kinda strange, and we don’t empathize with them, so very few people think that they are conscious.
But, it’s very hard to put ourselves in someone else’s shoes. In human understandings, as well, there’s a lot of misunderstandings and conflict that comes out of the inability to really experience things from another’s perspective.
But how would you ever demonstrate that a machine is conscious?"