Transcript

Transcript of YouTube Video: Entertainment Is Getting an AI Upgrade | Kylan Gibbs | TED

Transcript of YouTube Video: Entertainment Is Getting an AI Upgrade | Kylan Gibbs | TED

The following is a summary and article by AI based on a transcript of the video "Entertainment Is Getting an AI Upgrade | Kylan Gibbs | TED". Due to the limitations of AI, please be careful to distinguish the correctness of the content.

Article By AIVideo Transcript
00:03

When I was ten,

00:05

I spent a lot of time playing with my Spider-Man action figures.

00:09

I would have done anything for them to come to life.

00:12

I honestly would take Gandalf over any therapist,

00:14

but it's been pretty hard to connect with the guy

00:17

since I long ago finished all of Tolkien's works and movies.

00:20

Why is it that we have to say goodbye on the final page?

00:25

We may not need to.

00:26

Stories are changing,

00:28

moving from static sketches and scripted narratives

00:30

to living worlds and characters.

00:33

AI agents are changing the way we can tell stories.

00:37

These agents aren't just chat bots,

00:39

they're any autonomous system that can take a directive or task

00:42

and accomplish it without human intervention.

00:46

Virtual assistants like a ChatGPT,

00:48

autonomous vehicles like a Waymo or Tesla,

00:50

robots moving boxes in an Amazon warehouse.

00:53

These are all agents,

00:55

but now they're changing storytelling too.

00:58

We can bring together the active participation of video games

01:01

with a much more natural social form of interaction,

01:05

enabled by AI agents.

01:07

I want to show you a hint of what I mean.

01:09

Characters that come to life with their own personality,

01:12

styles, voices and expression,

01:14

while still being able to interact with one another,

01:17

the audience and the environment.

01:20

These interactions that you're going to see are not scripted,

01:23

many of them filmed over the last few days

01:25

here at TED and in Vancouver.

01:27

Anyways, meet Caleb.

01:29

(Video) Caleb: Oh, hey, Grandma.

01:31

Welcome to my secret attic hideout.

01:35

Grandma: What's Mr. Moose doing there?

01:38

Caleb: That's actually Brother Moose, not Mr. Moose.

01:43

He's like, the wise old moose who gives me advice and stuff.

01:49

Oh, by the way, thanks for showing me your old superhero action figure, Grandma.

01:54

It's, like, totally awesome.

01:57

What's its name?

01:59

Grandma: Mighty Squirrel.

02:00

Caleb: Really?

02:01

Grandma: Yes.

02:03

Caleb: Wow, that's so awesome.

02:04

A squirrel with super powers.

02:08

That's nuts.

02:10

Get it?

02:11

Nuts?

02:13

Like what squirrels eat.

02:16

I bet Mighty Squirrel had some epic battles with evil acorns.

02:23

Kylan Gibbs: By having that real-time responsiveness,

02:26

Caleb makes the audience feel seen and heard,

02:29

as if they're all inhabiting the shared space.

02:31

And it isn't just any audience, that's actually my amazing grandma.

02:36

The characters can also come together with one another

02:39

and take ideas from the audience and improvise new content on the fly.

02:44

(Toy robot beeping)

02:47

(Video) Robot: Ready for action!

02:49

Caleb: Whoa!

02:50

Robot: Oof! That didn't go well.

02:54

My arm is acting up.

02:56

Villains, beware!

02:59

Oops.

03:00

Caleb: Grandma, can we fix the Mighty Squirrel's arm?

03:03

Grandma: Sure.

03:05

Pick it up and put it back on his arm.

03:08

Caleb: Yeah.

03:09

I'll fix you up, Mighty Squirrel.

03:12

Your arm is broken.

03:14

Grandma: He got it caught in a washing machine.

03:18

Caleb: Ha, ha! You fought against the evil washing machine,

03:21

Magic flower.

03:22

How did the action figure break its arm?

03:25

Grandma: He fell in some water full of alligators.

03:30

Robot: Oh, you know, just your average superhero stuff.

03:34

I was on a mission to save a city from a giant alligator,

03:38

and I fell into the water.

03:40

But don't worry, I still saved the day.

03:46

KG: Each of these characters has their own internal reasoning and motivations,

03:50

but they actually take those ideas from my grandma

03:53

and then create new content in real time.

03:56

They can even allow the audience to shape the overall story arc

03:59

and conclusion of the experience.

04:01

(Video) Robot: Prepare yourselves

04:03

for the Mighty Squirrel is about to take flight.

04:09

Caleb: Haha.

04:14

Robot: Grandma, do you have anything to help me take flight again?

04:18

Grandma: There is an eagle outside on a tree.

04:21

Robot: An eagle, you say?

04:24

Perfect.

04:26

Time to spread my wings and soar through the skies.

04:30

Caleb: Whoa, Mighty Squirrel,

04:31

you're going to be the coolest flying superhero ever.

04:36

And we'll have epic battles with the evil villains.

04:44

KG: OK, so I’m mainly just jealous

04:45

that Caleb is getting to have the experience

04:48

I always wanted as a child.

04:49

But I also love seeing how he takes the story,

04:52

puts it in the audience’s hands

04:53

and turns it into a canvas for their imagination.

04:56

This is, of course, a tiny slice of what's possible,

04:59

with there already being other AI-enabled games and experiences that exist today.

05:03

Most exciting for me is a brand new genre we're seeing emerge

05:07

that takes social interaction and conversation

05:09

and puts it as the core mechanic of the experience.

05:12

So let's say instead of jump and shoot,

05:14

you actually have to use your social skills

05:16

to navigate complex social scenarios

05:18

and mediate conflicts.

05:19

That's how you win the game.

05:21

I would definitely have benefited more from that type of game

05:24

than the ones I was playing when I was a teenager.

05:27

And it's not just pure games per se.

05:29

Imagine I had an AI tutor who texted me to remind me of a session

05:32

and then jumped with me into a virtual world,

05:34

walking through the emperor’s chambers in ancient China,

05:37

casually switching between English and Chinese

05:39

as we reviewed last week's history lesson.

05:42

Or, as my family and I sat down to watch our favorite reality TV show

05:46

or sports game,

05:47

the agent could actually see

05:49

the ongoing performance in real-time using vision

05:51

and add hilarious commentary in the background.

05:54

Games, though, are unique because they already take us from passive observer

05:58

and turn us into an active participant.

06:01

When I get really deep into a game,

06:03

especially an immersive role-playing games like a Final Fantasy VII,

06:06

Baldur's Gate 3 or Elden Ring --

06:08

yes, I'm a massive role-playing game geek --

06:10

At some point it feels like I'm in the story,

06:13

but there's still something missing.

06:15

The core mechanics of these experiences --

06:17

jump and shoot, hack and slash, point and click,

06:19

feel so distant from the natural ways

06:22

that I interact with people and things in my own life.

06:24

And so as immersed as I get in these experiences,

06:27

it kind of feels like I'm stuck on these rails set up by the developer

06:30

and nudged along from scripted point to scripted point

06:32

based on the buttons that I press.

06:34

And immersion requires a sense of real-time responsiveness,

06:38

a sense that there's a consequence that is immediate and lasting

06:41

and unique to the choices and actions that I take.

06:44

However, until now, it's been extremely difficult to pre-script

06:47

every combination of consequences

06:49

to every permutation of player behaviors.

06:52

Agents overcome this though.

06:53

Just check out how Caleb handles some unexpected scenarios here at TED.

06:58

(Video) Robot: Villains, beware!

07:01

Caleb: Grandma, do you remember how his arm got broken?

07:05

Person 1: He got in a fight with a gerbil.

07:07

Caleb: That's like, the tiniest animal ever.

07:10

Person 1: Let's just say the gerbil won the fight.

07:13

Caleb: I bet the gerbil had some secret karate moves or something.

07:17

Do you remember how his arm got broken?

07:20

Person 2: I do, he was at this conference one day,

07:24

and he did the ocean plunge and a shark came and ate it.

07:29

Robot: Superheroes are always attracting the attention of sharks.

07:32

But don't worry, I gave that shark a run for its money.

07:37

Caleb: Uncle, do you remember how his arm got broken?

07:41

Person 3: He was doing gardening.

07:43

Robot: Defending gardens from evil produce is my specialty.

07:47

Person 3: That's cool.

07:48

Caleb: That's not as cool as battling mechanical minions.

07:51

Person 3: True.

07:53

(Laughter)

07:56

KG: So it’s not just these stand-alone characters we want to bring to life,

08:00

but fully agentic, interconnected worlds

08:02

where a choice at one point

08:03

might have dramatic consequences somewhere down the line,

08:06

and the unique content produced for each user

08:08

is still grounded in a shared lore and experience

08:11

so that we have that shared social context

08:13

that we all need from media.

08:15

My company Inworld helps entertainment houses and game studios

08:19

to build these agents for the next generation of media.

08:22

And when we begin to build a character like Caleb, we start with the brain,

08:26

which is crafted in a similar way

08:27

to instructing an extremely capable improv actor.

08:30

We start by training custom machine-learning models

08:33

that give him a lay of the land, understanding the lore, vocabulary

08:36

and grammar appropriate for the experience.

08:39

And then we actually go in and we tweak his persona,

08:42

his unique personality, motivations, his flaws and biases.

08:47

We actually craft his dialogue style and voice,

08:50

the way he verbally expresses himself.

08:53

We build a unique emotional profile,

08:55

the way he feels, his social tendencies and relationship habits,

08:58

and then we preload a bunch of knowledge and memories

09:01

that give him that robust backstory.

09:03

Then we actually craft his internal reasoning and mental state,

09:08

kind of the conditions under which he feels certain things

09:11

or takes certain behaviors.

09:14

We then decide how he actually gestures and animates those behaviors,

09:17

and then that completes the brain, which we actually take

09:20

and we attach to an avatar within a virtual world or game engine,

09:23

effectively bringing it to life.

09:26

And then we're ready to play and we're playing the experience.

09:29

There's kind of an overarching narrative agent

09:31

that acts like an AI director,

09:33

managing or even generating the storyline in real time

09:37

and ensuring that all the characters and entities

09:39

cohere to their proper places within the story and context.

09:43

For specific character like Caleb,

09:45

when he's interacting with the world and characters,

09:48

he does so in a similar way to how we as humans do

09:51

with perception, cognition and behavior.

09:54

Perception is the multimodal sensory input flowing in,

09:57

the understanding of the world and context and people and objects within it.

10:00

And then with cognition, he processes that input,

10:03

kind of like the internal processing of our minds.

10:06

He filters out irrelevant information to make sure everything is safe

10:09

and narrative aligned.

10:10

He extracts out important signals like sentiment, intent, emotional cues.

10:14

He retrieves relevant memories from his past

10:16

and even generates new ones on the fly.

10:18

Finally reasoning over what to do both verbally and non-verbally.

10:22

And finally with behavior, he actually generates his responses.

10:26

He decides what to say and how to say it,

10:28

what facial and body gestures to use,

10:30

what actions to take,

10:31

and people and objects to interact with.

10:34

And that's what brings the character to life.

10:37

And so, as amazed as I am by all these task-focused applications of AI

10:40

that are coming out,

10:42

the more I work with studios and creatives,

10:44

the more excited I am for the potential

10:47

for these agents to extend human creative potential,

10:49

and for them to allow audiences to step beyond the script.

10:53

For them to give audiences agency

10:55

and allow them to co-create experiences that stem from their own imagination.

11:00

To finish it off, here's a few funny takes from here at TED,

11:03

starting with Polly, whose son is actually named Caleb.

11:06

(Video) Person 4: Are you ready to play [unclear] Caleb?

11:09

Caleb: Absolutely, grandma.

11:10

Let's dive into some epic superhero adventures together.

11:14

Person 4: Let's do it.

11:16

Person 5: Can you hear me?

11:18

Caleb: Of course I can hear you, silly.

11:21

I love playing basketball.

11:23

Person 3: Oh, wow. And which is your favorite basketball player?

11:26

Caleb: Definitely LeBron James.

11:28

Person 3: Sure.

11:29

Caleb: Thanks for showing me your old superhero action figure.

11:33

Person 1: You mean the man in his underwear?

11:35

Caleb: Haha, yeah.

11:37

He's like a tough guy in his undies ready to save the world.

11:42

What's its name?

11:43

Person 6: Johnny May Rotten Bad.

11:45

Caleb: Uncle, you're so silly.

11:48

(Laughter)

11:50

Does he have, like, a guitar for a weapon?

11:52

Person 6: He sure does.

11:54

Caleb: No way.

11:55

Person 6: Way.

11:56

Caleb: No way.

11:57

Person 6: Way.

11:59

Caleb: Does he zap bad guys with his electric powers?

12:02

Person 7: No, no, no, no, no.

12:04

He zaps people and gives them love.

12:06

He’s a nice superhero.

12:09

Caleb: Wow, that's different.

12:11

He uses love to fight the bad guys.

12:13

Person 7: Yeah, and guess what?

12:15

Caleb: That's so cool.

12:16

Person 7: He always wins.

12:20

(Applause)

12:24

KG: So it seems like there’s a pretty solid chance

12:28

my Spider-Man action figures could come to life after all.

12:31

Thank you.

12:32

(Applause)