B1 中級 9 タグ追加 保存
動画の字幕をクリックしてすぐ単語の意味を調べられます!
単語帳読み込み中…
字幕の修正報告
Here's a joke my grandpa would have loved.
What do you call a murderer with moral fiber?
That's right.
A serial killer.
Sorry, I should have warned you.
Now I'm not proud to admit this, but there's a part of me that actually really loves this joke, too.
And I love it because it's so weird and so quintessentially human.
It's one of the last things we'd expect a computer to understand, and yet this joke was written by a computer.
That's right.
I'm talking about computational humor that's using computers to generate and understand humor.
It's an actual field, no joke.
So computers today see they're getting smarter.
They're getting smarter.
But they're also developing a sense of humor.
And as these algorithms of humor, so to speak, as they continue to develop, they have the potential to change how we relate to our mechanical friends, but also how we relate to each other.
And to be clear, I don't think this is just a curiosity.
As computers increasingly surround us in our lives, I think it's gonna be a necessity now.
I wasn't always convinced to the value of a relatable machines, let alone a making you laugh with softer.
And why would I need my software to like in the mood, right?
But then I took a closer look at myself, some of an angry man.
But I have routinely fantasized about taking my laptop and smashing it against Iraq, and it's always for the stupidest reasons.
Like maybe the Internet's a little slow is not even the laptops fault.
No people.
People frustrate me, too.
But the difference is that with people, I have a safety valve called humor.
Even on a call with Comcast, someone cracks a joke.
It changes the whole dynamic.
And this is because humor has a way of instantly connecting us with each other, even complete strangers.
You're more than 30 times is likely to laugh when you have company than when you're by yourself.
More broadly, we can look at humorous sort of the WD 40 of human interactions.
In a world where increasingly surrounded by computers, we're gonna desperately need some of that lubrication or we're going to drown in the frustration.
Now it's really important to recognize the significance of this problem.
Like computers today aren't just word processors.
They're on our wrists.
They're in our toasters.
There inside our medical devices there are becoming our our assistance.
Our companions even are caretakers.
And it's not just the movie her that I'm talking about here.
I mean Watson, Siri, Cortana, Google.
Now these these machines interact with people every day, and we're starting to see that humor can be helpful.
Siri, for instance, often has to deal with some pretty boneheaded questions like this one.
Siri, what color are your eyes?
Serious says in response.
I don't have eyes, but if I did, I think I'd be rolling them a lot.
Okay, that's great.
But now, even though this is clearly a can joke, I do think it hints at the ways that humor can reduce some of the friction between us and our machines, especially as they surround us.
But this is where computational humor gets really interesting.
It's not just about connecting us with our machines.
It can also help connect us with each other, especially for those of us who have trouble in social situations.
Now, in many ways, I I experienced this firsthand.
See, I wanted to be a comedian when I was a kid and there was a reason for that.
See, I I was an only child.
My parents moved around a lot, and the only constant friends I had really were in my computers or in the shows that I watched.
But that's not really a substitute for human connection.
So what?
Every new school in every new town, I sort of found myself back in the same place trying to make friends with strangers.
And for someone who shy, you're socially awkward.
That's terrifying.
So I decided I was gonna connect with people by making them laugh.
Now, this might sound like a fairy tale, but unfortunately, comedy is really hard, and I was really bad at it.
So I took what could very generously be described as an algorithmic approach to humor.
Okay, so I'd start with a joke like this one.
What do you call it be that eats too much Chevy?
Okay.
I was in elementary school.
Cut me some slack.
Now I'd recognize not being a total moron that the humor in this joke comes from the similarity of chubby and B.
And then I'd replicate this tons of times.
What do you call a B?
That's good for your health.
A vitamin B.
What do you call a newborn be a baby.
What do you call it being the springtime?
Maybe you get the idea.
No, you give going with us now.
What's what's interesting is e mean sometimes even this would be too much effort and I just copy the original joke out right.
You can probably see why I became an engineer now.
No, it's Ah.
What's interesting, though, is that this this process that describe for you this very hack ish and uncreative process It may not sound like an algorithm, but it iss In fact, at the same time that I was pulling this Carlos Mencia act in the early nineties, Kim been Sted Graham, Richie and a bunch of other researchers were busy jump starting the field of computational humor using what was essentially a very sophisticated ideas using the same fundamental ideas that I was using right and these stupid little examples that I shared with you.
Let me show you a simplified view of their process.
What do you call it when dinosaurs collide?
Trying to source Rex?
Okay, that's our fun laugh.
There's four words here.
Dinosaurs collide trying to source Rex those There are key words and they relate to each other in very specific ways, both in terms of how they sound.
Some of them sound like and in terms of what they mean some of the mean similar things.
So we can ask a computer to go hunt through dictionaries and corpora for other combinations of words and phrases that have a similar pattern of relationships right and what comes back.
It's usually pretty awful, but every once in a while we get something that my grandpa would approve off.
For instance, this one.
What do you call a material with the bright side, a silver linen?
So again you see, there's four words here, and they relate to each other in a very similar way as the previous example.
So this is a lot of fun, but it's also more than that Justus, in my example.
This, too, can help connect people with each other.
In the mid two thousand's, a group of researchers from Aberdeen, Edinburgh and Dundee created a version of the system called Standup, and the goal of the system was to help kids who have communication disabilities of some sort.
Now, a lot of what we call humor, we acquire through interacting with each other and participating in wordplay.
And sadly, some kids don't have very many opportunities for that.
And that's where the software comes in.
It gives them sort of a playground where they can play around with puns, generating puns and modifying the generation of puns as well.
You could actually play with it a version of it yourself.
It's called the Joking Computer.
Now it's admittedly a focused example, but I think there's something more profound happening here.
It's not unusual for those of us who have trouble interacting with each other to go to computers, a sort of a safe place, a safe haven.
There's something really powerful about taking that same space and using it to teach us how to laugh and how to make other people laugh.
I find that really beautiful, but to use this more generally and broadly, we're gonna have to go beyond puns to the more unstructured and subtle humor that we humans and engage in pretty regularly.
I mean, think about the last thing that made you laugh.
Chances are, not only was it not a pun, it probably wasn't even a joke.
In some sense, the real goal for us here.
It's not necessarily to create machines that are going to write jokes for us, but to create machines with personalities that we find humorous or amusing.
Now.
To get two personalities, though, we often have to go through language and language is a bear.
Your average English speaker knows tens of thousands of words and breaks grammatical rules about as often as he follows them.
And even if you can get past that, there's issues of ambiguity, context and general common sense knowledge.
When I ask you, how much does President Obama make?
Somehow, you know, I'm asking about his salary, not about how much soupy mix.
This is very hard to encode into an algorithm, but in recent days we may have caught a break.
We may have found a back door, and this back door has a big sign on it.
That's a hint, and the sign reads data.
The amount of data specifically human generated data that's available to us today is unprecedented, and it's growing.
And the reason it's so valuable.
One of the many reasons I should say is that with it we can teach our machines how to communicate with us.
You've actually experienced this yourself.
If you've used an Internet search engine in recent days like Google being duck taco, what have you?
These machines have been trained by data, how to understand your queries and your questions.
But data could also go a step further and teach machines howto have him using personalities.
My favorite example of this actually happened earlier this year, and interestingly, I don't think it's typically seen as an example of computational humor.
But it ISS so.
Several months ago, a graduate student at Stanford named Andre Car Poppy introduced a piece of software called Char R nn.
Now what the software does, like some other pieces of software algorithms out there is it is it trains machines to imitate a data set of your choice.
So, for instance, I compiled a very large list of tweets from believers, thes air fans of Justin Bieber and I fed it to this machine.
This machine digested all these examples and gave me back a trained machine that could generate fake believer tweets.
This clay Yeah, basically imitating the style of this data set.
All right, so when I ask this machine to talk, here's the kinds of things that says I love the way he is so beautiful, baby and hashtag MTV hottest.
They love you so much crying, emerging.
Okay, so I'm not an expert on believers, but to my untrained eye, this seems like a pretty good imitation of the real thing right now.
What's interesting is, in recent days, people have been trying this with all sorts of data, political speeches.
The Bible, 50 Shades of Grey is actually one of my favorite examples.
I would apologize for all the white out, but honestly, you're not missing much.
All right now the point.
I'm the reason I'm showing you all these examples This.
I want you to notice something.
In none of these examples is the machine cracking a joke.
It's not running a standup routine.
It's just the humor is a lot more settled that's buried inside the style of language.
You could say that we've painted this machine's personality with data, and I think that's a step in the right direction.
But it's also just the tip of the iceberg.
There are so many ways in which computational humor can take advantage of data.
Today, researchers have built irony detectors installed generators.
Even, uh, even there's even, uh, there's even a double on Tonda.
Recognize her out there called deviants.
That's actually the acronym for the algorithm.
No, we're also just really at the very beginning with these algorithms.
There's so much about humor that they're not even remotely capturing yet to take things forward.
The timing couldn't be better on On one hand, we have this incredible tool called data that's getting more powerful by the day.
And on the other hand, our computers air surrounding us more and more every day, and and the need for them to be more personable and relatable, that's just gonna get stronger.
I think computational humor could be the key to making healthy relationships between us and our machines and also less frustrating lives for us.
And in the process, it might even connect a few of us Some may disagree.
To Quote and Wilson Shafe, I recognize that humor isn't for everyone.
It's only for those of us who want to have fun, enjoy life and feel alive.
And here's to the algorithms that are helping us get there.
コツ:単語をクリックしてすぐ意味を調べられます!

読み込み中…

Machines need an algorithm for humor. This is what it looks like | Vinith Misra | TED Institute

林宜悉 2020 年 3 月 20 日 に公開
お勧め動画
  1. 1. クリック一つで単語を検索

    右側のスプリクトの単語をクリックするだけで即座に意味が検索できます。

  2. 2. リピート機能

    クリックするだけで同じフレーズを何回もリピート可能!

  3. 3. ショートカット

    キーボードショートカットを使うことによって勉強の効率を上げることが出来ます。

  4. 4. 字幕の表示/非表示

    日・英のボタンをクリックすることで自由に字幕のオンオフを切り替えられます。

  5. 5. 動画をブログ等でシェア

    コードを貼り付けてVoiceTubeの動画再生プレーヤーをブログ等でシェアすることが出来ます!

  6. 6. 全画面再生

    左側の矢印をクリックすることで全画面で再生できるようになります。

  1. クイズ付き動画

    リスニングクイズに挑戦!

  1. クリックしてメモを表示

  1. UrbanDictionary 俚語字典整合查詢。一般字典查詢不到你滿意的解譯,不妨使用「俚語字典」,或許會讓你有滿意的答案喔