Hello again 欢迎你又回到酒馆。

Hello everyone, and welcome back to geek time. Hello, Brad.

Hello.

Last time we talked about aliens and that thoroughly freaked me out. And today I thought we talk about another really, really big and really popular topic, AI.

Yes.

First of all, when it comes to AI - Artificial Intelligence, I think for most people, probably our ideas about AI are mostly from movies. Are there any quintessential AI films that you can think of?

I think from the most part, people think about the movie AI or Short circuit, Chappie and even the Terminator.

Okay, first of all AI,人工智能, that was about a little boy. Right?

Right.

It's actually more on the emotional side.

Right.

What about Short Circuit? 霹雳五号

It's about a robot who is built by the military. But he gets hit by lightning and become sentient. He basically starts learning about the world and he becomes alive.

And what about Chappie, 超能查派。

Yeah. So Chappie is very similar to Short Circuit. This one is a more modern tale. Developer of the robots wants to build something a little bit more functional. And so he designs a learning robot. So he implants this new AI into one of the robots from his work. The movie starts off with the robot being very childlike, but growing over time with its experience.

So almost like a a real human being.

Right.

And I've noticed that youb’ve used the word sentient, s e n t i e n t it's also pronounced /'senʃnt/ I've heard both pronunciation. It basically means if you say a robot is becoming sentient, it means it's capable of feeling emotions. Right? Is that what sentient means? They're conscious.

Essentially, yes.

Which reminds me of Westworld. Is these AI hosts they developed sentience. By the way. You've mentioned all these movies AI, in the movie AI the robot boy is more emotional almost just like a little boy. So that's more like nice side of sentience, and then Short Circuit and Chappie, they just sound like comical characters. They're like fun and generally harmless. But then you have things like Terminator.

So Terminator is basically kind of like this futuristic AI that decides that it wants to eradicate humans because it sees humans as a threat so it sends back a robot from the future to destroy the leader of the resistance in the future.

That does give you nightmares, not just some of the graphics, but it's also the whole idea of sentient robots, when AI develop sentience and then they are going to destroy humanity. But having said that, nowadays AI for most people, it hasn't developed to that level yet.

Right.

So what are some of the applications of AI that we can actually see in our day to day life?

We see a bunch of things like chat bots. They're basically just kind of like these rudimentary AIs. They're not really thinking in the sense, but they're kind of a start.

Chat bots聊天机器人, for example, like iphone ’s Siri?

Yeah. So they have like older ones where you send it a message and it sends response. But now they have like verbal ones like Siri or Cortana where you speak to it and it responds to you with a voice.

When you ask Siri more sort of random questions, not just information based, how did they collect those data?

So basically, they looked through a huge corpus of Information. Basically, they have lists of conversations that people have had. And they go through the recordings and listen for a similar question. And then they give a response based on the responses that real people have given.

Is that why sometimes the conversations are a bit strange? The conversations that I had with my Siri?

图片

Yeah, like they're not listening to you specifically. They are just kind of looking at your question and then looking on the internet for an answer and then giving you that answer.

I tell you what to make it fun. Let's try. I have my iPhone with me. I'm gonna try some fun questions with Siri and see how Siri respond. Let me ask Siri. First of all,

Hi, Siri.What phone is the best?

The one you're holding?

And what does the fox say?

Ring-ding-ding-ding-dingeringeding!

What does that even mean, Brad?

It's from a funny song that came out a few years ago. It's one of the sounds that the fox says during the song.

So it's not just generated out of random. It's still like you said based on statistics based on data it’s been collected. Ok. And then let me ask ... Hi, Siri. Read me a Haiku.

Haiku can be fun, but sometimes they don't make sense,Hippopotamus.

I suppose Hippopotamus, that is five syllables.

Yeah.

Ok. And then questions like, hi, Siri, what is the meaning of life?

42.

What? What's the meaning of life? And it's 42. What does that even mean?

If you know there's a book called the Hitchhikers Guide to the Galaxy《银河系漫游指南》And that's kind of a joke from the book. So probably whoever is programming the responses to that decided to put in that as a joke for people who have read.

So these programmers, they're probably putting in a little bit like geeky jokes in it. right? One more. What's your favorite animal?

Software doesn't usually get to

choose one, but I'll say giraffes,

what's yours?

T Rex.

Wow, your favorite animal lived

millions of years ago,

I like your style.

Okay? But joke aside, these are some of the weird conversations. Sometimes I get bored, I talk to Siri, but I find that it's difficult for her to really keep the conversation going at certain points. And she will get stuck. And she would say things like, I don't know, I don't understand what you're saying, or I don't really know how to answer that. So it's not really a conversation in that sense.

It's not really a conversation. Like, when we're talking, we can keep the thread. We know what happened before in the conversation, but a chat bot usually just analyzes the last response or the last question that it received. And so it can't really go beyond that. And so if you're trying to have a long conversation with it, you're gonna lose it in after one or two responses.

So there's no string of thoughts.

Right. The chat bots, they learn, they do have some minimal learning. There was one chat bot that Microsoft made a few years back. A lot of people were making these responses and some of people were saying pretty racist things. So as the chat bot was building up responses, it started responding to people with very racist comments. So it had to be taken down. But basically it was just a bunch of people playing around. And this taught the AI basically how to be a racist.

I mean it's pretty scary if you think about it, if how they learn is just by collecting all the answers from people, what if the answers are jokey ones or just evil ones or just like you said racist ones, and then it will learn based on that.

Right

Now coming back to the main topic AI, exactly what is AI?

There's different levels of AI, and AI is something that thinks on its own. A narrow definition or a weak definition of AI is an AI that we can do now. And it's basically something that works on one task, like a chat bot can chat, other types of AI that might do one simple task, like driving a car. And it learns how to drive through the streets, things like that.

It's like a singular task.

Right.

You have a weak one, then you definitely have a strong one. Right?

Yeah. So we have the general or strong AI. Now this AI can think like a human. But it doesn't have emotions like a human.

So not sentient.

Right. It's not sentient, but it can kind of understand our emotions. So it can recognize the fact that we are having emotions. So when it sees people crying, it will know that something is going wrong. In the movie Terminator, for example, the robot sees the boy crying and it can understand that the boy has feelings, but they can't really make their own. They might be able to mimic them, but they don't really have their own feelings.

They don't feel it. Actually talking about feelings. I'm gonna get Siri again. Hi, Siri, can you laugh?

Ha ha!

I guess that's what you're saying. Right? She doesn't feel happiness. She just knows laugh is ha ha!

So we have weak AI, and strong AI. There is also something called artificial super intelligence. Right? What is that about?

That's basically it's an AI that does have emotion, it can feel emotions. But beyond that it can be much smarter than humans. It can go beyond us. So as far as our intelligence, but not only just our intelligence, it also has the emotional capability,

That would be all these things. They have been depicting in si-fi films. That's gonna be the end of humans I guess.

Possibly.

If you watch Westworld, that's going to bring you some fear about Artificial Super Intelligence.

Yeah.

Another， I think， topic that's quite related to AI is IOT - internet of things. 物联网, what is that ?

So basically, we've created this huge combination of devices and processing power around the world. We have our Alexa, we have other devices that connect to our fridges our toasters, our computers. All these devices in your house are connected to the internet. That internet could become a brain, it could become sentient at some point. There could be a singularity which creates a brain that combines everything on the internet.

But that would be really scary if there is a main AI brain that's controlling everything in our homes.

Yeah. Like, if you have something that's controlling everything around your house, what happens when it goes on the fritz? You'll get stuck in your house. You wouldn't be able to unlock your door because it might control the locks on your door. It may control the air conditioning, the refrigeration, and all your food could go bad. Even the heating and the water, you might not be able to take a shower.

So the more smart devices we use, actually we are potentially putting ourselves in more danger.All right. We're gonna wrap up here. And these are some of the basic concepts and basic levels and definitions of AI. In the advanced episode, how about we get more into the future of AI? What if they do develop super fast?

Yeah, the next episode will definitely be very interesting.

And more disturbing hopefully.

Very disturbing.

Alright, see you next time. Thanks Brad.

No problem. Thank you. Bye bye.