Home
เข้าสู่ระบบ
สมัครสมาชิก
Loading...
ฝึกฟัง
ฝึกฟัง
/
Video
/
TED Talk
/
How to Raise Kids Who Question AI | Randi Williams | TED
How to Raise Kids Who Question AI | Randi Williams | TED
เลือกโหมดการเรียน:
ดูคำบรรยาย
เลือกคำ
เขียนคำใหม่
Highlight:
3000 Oxford Words
4000 IELTS Words
5000 Oxford Words
3000 Common Words
1000 TOEIC Words
5000 TOEFL Words
คำบรรยาย (280)
0:08
In the 1940s,
0:10
the radio was a transformative technology.
0:14
But it wasn't an immediate hit.
0:16
That's because the radios of the early 1900s looked like this:
0:21
unwieldy contraptions of exposed wires and glowing vacuum tubes,
0:26
like a science-fair project gone wrong.
0:28
It wasn't until 1929 when radio started to look like this,
0:32
with its sleek Bakelite --
0:34
that’s a kind of plastic shell --
0:36
that they became insanely popular --
0:39
but at a critical cost.
0:42
Transparency.
0:44
What happens when technology becomes a black box
0:48
and we stop being able to see what's going on inside?
0:52
Fast forward 100 years.
0:54
Today, the sci-fi artificial intelligence devices of our dreams
0:58
have been woven into the very fabric of modern life.
1:02
But like the radio, these systems are “bakelited.”
1:06
We've hidden the machine behind frictionless interfaces,
1:09
beneath friendly voices
1:11
and inside children's toys.
1:13
Over the past decade,
1:14
I've had a lot of conversations with kids about AI
1:18
as an MIT researcher and founding member of Day of AI,
1:21
a nonprofit creating opportunities for millions of children around the world
1:25
to learn about how AI works.
1:29
And here's what I've been seeing.
1:31
These devices mimic intelligence and friendliness so well
1:35
that children learn to trust them --
1:38
sometimes more than they trust themselves.
1:40
And there is real danger of children becoming overly reliant on
1:44
or inappropriately attached to their smart toys.
1:48
That said, with some intentional choices on the part of adults,
1:51
we can help children see the technology for what it really is,
1:55
objects that they can play with, or even program,
1:58
rather than entities that they should look to for all the answers.
2:02
My journey exploring kids
2:04
and their relationships with AI
2:05
began early in my days at MIT,
2:07
working on this fluffy, adorable social robot, Tega.
2:13
So Tega being a social robot means
2:15
that its goal is to help people with their goals.
2:19
Social robots are used throughout healthcare,
2:21
customer service and education.
2:23
In Tega's case,
2:24
we were building it to see if a robot could help young children
2:28
as they were learning how to read.
2:29
Let me show you a quick clip of Tega in action.
2:33
Student: Hm, I'm lavender.
2:37
Tega: What are we trying to find?
2:39
Student: We are trying to find lavender-color stuff.
2:44
Tega: OK.
2:47
The word was lavender?
2:50
Student: Yes.
2:52
And what is lavender, though?
2:57
Tega: Lavender is purple.
3:00
Student: Yes.
3:02
Just like my friend loves lavender.
3:06
Tega: Hmm.
3:10
Randi Williams: So as you saw,
3:12
it's Tega's job to keep the child engaged in the story.
3:16
Tega will ask questions, add funny commentary,
3:19
and of course, weave in that target vocabulary for the week.
3:22
And it was my job to introduce Tega
3:24
and get children oriented to how to use it.
3:27
So one day I come into the study room and I say, "Hi,
3:30
my name is Randi, this is Tega.
3:33
You're going to read some stories with Tega today.
3:35
Let me show you how it works."
3:36
And the girl I was working with, a kindergartener, stops me.
3:40
"No thanks," she says.
3:42
"I know all about robots.
3:44
I have an Alexa at home, and she's my best friend.
3:46
Watch this."
3:48
And then she proceeded to show me how my robot worked.
3:51
(Laughter)
3:52
What do we make of a response like that?
3:54
Well first, I thought it was kind of impressive
3:57
that this girl who, again, was learning how to read,
4:01
could effortlessly navigate the internet
4:03
thanks to voice technology.
4:05
And for context, I had to know how to read, write and type
4:09
before I could ask Jeeves a question.
4:12
But for her, not so much.
4:14
But secondly, I was a little bit disappointed
4:16
that this girl didn't think that my super cool robot was so super cool.
4:20
And when I took a step back, I realized it wasn't just her.
4:25
We are raising a generation of children who are growing up with AI
4:28
and smart toys,
4:30
and they tend to see these devices not as virtual assistants or technology,
4:34
but as friends,
4:36
distinct beings with a mind, a will and a soul.
4:40
And that made me really curious
4:42
about what children make of this technology.
4:45
How much do they understand about what's going on?
4:48
So to investigate that question,
4:50
I ran a study.
4:51
I had 30 children between the ages of three and 11
4:55
come together to interact
4:56
with a whole bunch of different smart toys all at once.
4:59
We had toy robots, talking dolls, text-based chatbots,
5:03
and of course, plenty of virtual assistants
5:05
like Alexa and Google Home.
5:08
One of the girls in the study, her favorite animal
5:11
was sloths.
5:12
She loves sloths.
5:14
She'd been telling us all day how much she loves sloths.
5:17
And so naturally, when she got hold of an Alexa,
5:20
she wanted to learn more about sloths from Alexa.
5:23
So she picks up Alexa and says,
5:25
"Alexa, what do sloths eat?"
5:28
And then Alexa, very helpfully, replied,
5:30
"I'm sorry, I don't know how to help you with that."
5:34
The girl looked disappointed for a second
5:36
and then all of a sudden perked right back up.
5:38
"Oh, I know," she says, and she picked up a second Alexa.
5:42
"I'll see if this one knows."
5:43
(Laughter)
5:45
The limitations of AI are not only frustrating and confusing for children,
5:50
but research in child-robot interaction shows that it can make them vulnerable
5:54
to even more serious risks,
5:56
like manipulation or accidental exposure to inappropriate content
6:00
or security breaches.
6:03
And for me personally,
6:05
even the devices that are marketed as educational
6:08
or screen-free are still corrupt because they're consumer products.
6:13
And they seem to be designed as something
6:16
that can be endlessly engaging or addictive.
6:19
They want to become your child's best friend and gain their trust
6:23
so that they can advertise to them.
6:25
Or better yet,
6:26
sell that trust back to you for $20 a month.
6:30
The bottom line is that smart toys open a digital door
6:34
into children's playtime, bedroom and hearts.
6:37
And if you're thinking about opening that door,
6:40
my advice to you would be to take a very careful look
6:43
at what's on the other side.
6:46
The good news is that we can mitigate the risks of AI to children
6:51
by leaning into something they already do.
6:54
Kids are natural reverse engineers.
6:57
Over and over again,
6:58
I've watched children playing with AI devices
7:01
trying to break it,
7:03
just asking seemingly absurd questions like,
7:06
"Alexa, how old are you?"
7:09
Or “Cosmo” -- which is a toy robot car --
7:12
"Can you jump?"
7:13
Or "Hey Google, is it OK if I eat you?"
7:17
(Laughter)
7:18
And these questions, while silly,
7:21
also are an example of children probing the machine
7:23
about its very nature.
7:25
Play is inherent to how children understand the world,
7:29
which means that maybe we can use it to help them understand what's happening
7:33
with the technologies they're interacting with.
7:36
What if children could take apart
7:39
and reconstruct their smart toys
7:41
the same way that they take apart and reconstruct physical blocks?
7:45
That was the idea behind another social robot that I worked on, “PopBot.”
7:50
As you can see, PopBot’s body is made completely out of LEGO bricks,
7:55
and it's kind of like a typical smart toy,
7:57
except I built it to make the most complex ideas of modern AI --
8:01
logical reasoning, machine learning, generative AI --
8:05
easy to understand through hands-on, child-driven play.
8:10
The idea was, if a four-year-old could understand AI,
8:13
then anyone can.
8:16
So for example, to learn about machine learning,
8:18
what children would do is they would play Rock, Paper, Scissors against the robot.
8:23
Let me show you a quick clip of what that looks like with PopBots.
8:26
Person: OK, go ahead.
8:27
Student: Scissors... shoot.
8:29
I choose rock.
8:31
PopBot: Rock.
8:33
You chose rock, and I choose rock.
8:36
PopBot: It's a tie.
8:38
Student: Rock!
8:39
Student 2: It's a tie!
8:42
RW: The beauties of a real-life classroom study.
8:45
So before children could start playing against their PopBot,
8:50
first they had to teach their PopBot the rules of Rock, Paper, Scissors.
8:53
And then, as they played the game,
8:55
rounds like that, over and over again,
8:57
what PopBot would try to do is pick up on their pattern
9:01
so that it could predict what move it could play to beat them.
9:04
And if you're between the ages of four and six,
9:06
pretty soon it gets hard, if not impossible, to beat your PopBot.
9:10
There was one boy who was very frustrated by this,
9:14
and so I planted a mischievous idea in his head.
9:17
I said, "Well, what if you taught your PopBot
9:20
the wrong rules of Rock, Paper, Scissors?
9:22
That way as you played it, you would win every single time."
9:26
(Laughter)
9:27
And he really liked that idea.
9:28
(Laughter)
9:30
Our little act of AI sabotage created a unique opportunity
9:33
to talk about who teaches AI the rules.
9:36
Well, people do.
9:38
And so I asked him, "Well,
9:39
if you can teach your PopBot the wrong rules of Rock, Paper, Scissors,
9:43
how do you know Alexa was taught the right ones?"
9:48
Which led to a very insightful conversation where he discussed
9:51
how much we can trust AI
9:53
and the importance of verifying what it says, with an adult’s help,
9:56
just to make sure that Alexa has been taught all the right rules.
10:01
The big idea that I want children to understand is that AI is not magic.
10:08
It's a set of rules written by people.
10:11
And it's up to us,
10:12
those of us who have been around for a bit longer,
10:15
to make sure that children aren't being programmed by their toys,
10:18
but rather things work the other way around.
10:21
We need to raise a generation of children who know that they are the ones
10:25
who get to write, and even rewrite, the rules of AI.
10:29
And this work starts at the dinner table.
10:33
So you're all sitting down and a question gets asked to the nearest AI chatbot
10:37
or smart toy.
10:39
When it responds, rather than just accepting that answer,
10:43
wonder, "I wonder why it answered the way it did."
10:48
Or ask aloud,
10:49
"Do we agree or disagree with that answer?"
10:53
"Who do we think wrote the rules to make it work the way that it does?"
10:58
And the thing is,
10:59
you don't need to know the answers to all these questions
11:02
or become some sort of AI expert --
11:03
that’s not the point.
11:05
What children need is a model
11:07
of what it means to be a curious user of AI.
11:10
They need someone who will sit down with them,
11:13
explore the machine,
11:14
poke at its limits,
11:15
challenge its responses,
11:17
and most importantly,
11:18
dare to rewrite its rules.
11:21
And I've been thinking about this a lot in the context of what's happening
11:25
in children's toy boxes,
11:26
but the bigger picture is
11:27
that we should all be treating the everyday AI systems
11:30
that we interact with the same way.
11:33
I 100 percent believe that we can help children build safer relationships
11:38
with their smart toys,
11:39
and we all need to build better relationships between ourselves
11:43
and the technologies we interact with every day.
11:46
I believe in a world where no one is left in the dark by technology,
11:50
and where everyone has the inherent right to play a part
11:55
in shaping the systems that shape our lives.
11:58
Thank you.
11:59
(Cheers and applause)
12:06
Chris Duffy: Randi, fantastic talk.
12:08
I'd love to ask you just one question here, which is,
12:12
before I was a parent,
12:13
I had some very black-and-white thinking around screens.
12:17
I had this pristine vision of, I would never use screens with my child at all.
12:21
And then as I think many parents have experienced,
12:23
that crashed into the reality of needing to put sunscreen on him
12:26
or to clip his nails.
12:28
I'm wondering, AI, as a new technology
12:31
and these smart toys,
12:33
how do you see parents balancing
12:36
the ideals and the best case of what we want
12:39
versus the practical realities of kids being exposed to these things
12:42
in ways that are sometimes out of our control, even as a parent?
12:45
RW: Yeah, I really love that question.
12:48
I think, first, because it gives so much grace to the fact
12:53
that parenting is hard
12:54
and there are a lot of challenges.
12:56
I think, in the world of artificial intelligence,
12:58
something I think a lot about is human wisdom, right?
13:02
And we've actually been kind of struggling with this problem for a long time.
13:05
There were books and then the radio and then TV and then the internet,
13:10
and now it's artificial intelligence.
13:13
And I think the advice from then still rings true now,
13:16
which is, don't leave children on their own to figure it out.
13:19
Rules are great,
13:20
but better is to help your child learn how to interact with them
13:24
in a way that works with your family's values
13:27
and the ways that you want to engage in the world.
13:29
So if you're willing to take that journey with your child,
13:32
I think that could do a lot of good.
13:34
CD: Thank you, that's a great answer.
13:36
Another big round of applause for Randi.
13:38
(Applause)
