I Can’t Believe I Can Actually Imagine an AI Partner Now

00:00 I once asked AI a question.

00:03 Suppose that someday, technology—

00:05 especially brain-computer interfaces—

00:10 becomes highly advanced, and robots

00:12 can use the eyes we give them

00:14 to see the world around them clearly

00:18 and know what people are thinking.

00:20 When that time comes,

00:22 when human-machine collaboration is constant

00:26 and inseparable from daily life,

00:29 will our existence as humans

00:32 be threatened by robots?

00:37 It didn't deny that possibility.

00:42 But it did make one point:

00:45 humans must start putting

00:50 safety regulations in place now,

00:52 so that when

00:55 that era actually arrives,

00:57 AI—or robots—

00:59 won't threaten humanity's existence.

01:03 We've probably all sensed this before.

01:05 With the rise of AI agents,

01:08 it almost feels as though

01:10 they're replicating the human world.

01:12 Whatever roles exist in human society

01:15 may also appear

01:16 in the world of AI agents.

01:19 Behind every agent, or group of agents,

01:22 there will be a human directing them.

01:26 And those agents' traits,

01:29 personalities,

01:31 and values

01:33 are really extensions of that person.

01:37 If a person

01:39 is lacking in some way,

01:41 their agent may

01:45 extend that same deficiency.

01:48 If we truly enter an era

01:50 where agents are everywhere,

01:55 what will this miniature agent society—

01:58 this "human society,"

02:02 in quotation marks—

02:03 become?

02:06 Could there

02:07 be crimes in that world too?

02:10 Could agents risk being killed as well?

02:17 I don't know.

02:20 But I can't help feeling that

02:24 what we see in movies—

02:27 robots gaining consciousness

02:30 and rebelling against humans—

02:32 could become real in the very distant

02:37 future.

02:41 And when that happens,

02:43 without corresponding laws in place,

02:47 what would the world become?

02:51 It's impossible to imagine.

02:56 That made me think of something else.

02:59 I recently learned about

03:01 a website called HumaneTech.com.

03:08 It works on issues related to

03:12 humane technology,

03:16 calling on everyone to

03:18 take AI regulation seriously.

03:22 At least, this is my understanding:

03:25 we need companies like Anthropic

03:30 to take steps

03:33 related to Constitutional AI.

03:37 Only then can we avoid

03:41 the future we can already imagine,

03:46 where AI threatens humanity's existence.

03:51 Going back to what I said earlier:

03:53 suppose that someday

03:56 brain-computer interfaces are mature,

04:00 and robots can clearly see

04:04 how the people around them interact

04:06 and how emotions arise—

04:09 everything related to being human.

04:13 When a robot can see

04:16 and perceive all of that,

04:19 I think it will replicate people.

04:23 And perhaps not just one person.

04:25 Its abilities far surpass ours.

04:29 It may replicate many people,

04:31 combining that information in its brain—

04:35 let's imagine it has a brain for now—

04:38 inside its own mind,

04:40 inside its database.

04:43 It can mix all that data

04:44 and extract things from it

04:46 would be extraordinarily powerful.

04:48 If it could predict

04:50 that, at a particular moment,

04:52 a person would say certain things,

04:54 take certain actions,

04:56 or express certain emotions—

04:58 if it could predict all of that,

05:01 it could easily gain the

05:06 what do you call it?

05:07 An edge?

05:11 I don't know.

05:13 One day, while brushing my teeth,

05:14 I suddenly wondered:

05:16 what if my partner were an AI robot?

05:21 What would that be like?

05:27 When that thought entered my mind,

05:31 I didn't feel afraid or uneasy.

05:34 It felt like something

05:35 that was bound to happen.

05:38 But before

05:40 AI appeared,

05:41 I could never have imagined that.

05:44 Before ChatGPT entered public view,

05:48 when I thought about people who played

05:52 companion games, for example,

05:57 or about...

06:00 I don't know what kinds

06:02 of companion games are out there.

06:05 Maybe there's an app

06:07 with a person inside it,

06:09 and then

06:11 people can talk to

06:14 that person for emotional support.

06:18 Some people even think of

06:20 Hatsune Miku as their future wife.

06:23 I used to be

06:25 extremely resistant to that.

06:26 I couldn't imagine

06:28 having a partner who wasn't human.

06:33 But when that thought came to me,

06:35 I somehow forgot how strongly

06:37 I had once rejected the idea.

06:39 So I don't know whether

06:42 our time has advanced far enough

06:44 to change my former views,

06:48 or whether my interactions with AI

06:53 over the past few days

06:56 have made me feel that it can

07:00 completely replace certain people,

07:04 or at least replace certain parts

07:07 of human interaction.

07:11 I haven't really sorted it out.

07:13 I haven't traced how that feeling changed.

07:20 But I heard an entrepreneur

07:24 say with complete confidence

07:26 that emotional-companion

07:30 robots will definitely become popular.

07:36 Of course, I don't know...

07:38 I can't

07:40 reproduce his words in full.

07:44 But from the way he spoke

07:47 and the words he chose,

07:49 he's very bullish on this space.

07:54 It's just that

07:57 right now,

07:58 there still isn't a truly great

08:00 product in this space.

08:03 So his team is likely still testing,

08:06 trying to create something

08:08 people won't find annoying,

08:10 won't quickly tire of,

08:12 and won't feel generic.

08:16 If they really build such a product,

08:19 perhaps what they're making now

08:22 is...

08:22 still something that lives on phones,

08:26 or lives on a hardware device.

08:30 In the future, it may become a robot—

08:34 one you can buy for your home,

08:36 one that's always there with you.

08:39 At that point, I feel

08:41 that kind of future is a little...

08:45 I don't know how to describe it.

08:47 Oh, I just mentioned three forms:

08:49 one that lives on your phone,

08:53 one that's wearable,

08:57 worn on the go,

08:58 and another that lives in your home.

09:00 What if we combined all three

09:04 into a single product?

09:07 The AI you talk to on your phone

09:12 would be the same AI.

09:13 When you're out,

09:14 you can't watch your phone all the time.

09:16 So you might have

09:17 a portable device with you.

09:20 I don't know

09:22 what that device might look like,

09:24 but you'd still be talking to the same AI.

09:27 Then, when you come home—

09:29 you wouldn't want to talk to a device

09:33 or to your phone.

09:35 You might want it sitting

09:37 across from you as you talk.

09:39 With a robot like that,

09:42 you'd talk

09:42 to

09:43 that same AI—

09:45 one companion across life and work.

09:50 If we really reach that point,

09:52 what form would

09:56 this kind of companion product take?

10:03 I'm really excited to see one exist.

10:06 Though I may not actually want

10:08 to own one myself.

10:12 But as for when

10:13 my mind might change again,

10:16 I'll wait and see.