00:00¿Qué pasa?
00:33People, at least so far, tend to be much better at unexpected situations and being able to
00:40reason about something that you've never seen before, but be able to figure out on the fly
00:47what to do, especially if you're using a machine learning-based system. It relies critically on
00:54the training data that's been given, and if you come up with a completely what's known as out-of-distribution,
01:00which is sort of a code word for just new or novel situation, then it's going to be much
01:08more difficult for a software-based system to react than people are.
01:19It's partly the autonomous car side. It's partly also the human expectations.
01:24Just in the same way if you have an Alexa at home, you realize the first time that when
01:30you say something it doesn't understand you, then you figure out how to talk to it. You
01:33change your behavior in response to the technology. I think that will also happen with autonomous
01:38cars, that the people will change their behavior once they have a good model of what they do
01:43and don't understand.
01:56I feel like a human may be better at creative thinking and stuff in situations where normal
02:04road laws don't apply, like if there's a block up on the highway or if it's the marathon coming up
02:09and
02:09like certain roads are closed off, a human might be able to like better navigate new situations
02:15versus a robot where it's kind of just relying on programming.
02:18So you're there.
02:18It's a theme, right?
02:19They said, I have fallen for a co-worker and I think he has two because he tells me the
02:25sweetest things.
02:27Just?
02:27Now we go.
02:37Gracias por ver el video.
Comentarios