Did We Really Only Start Outsourcing Taste and Judgment in the AI Era?

00:00 I've long wanted to discuss

00:03 aesthetics

00:05 and outsourcing judgment.

00:09 Many say that in the AI era,

00:13 the most important things are aesthetics and judgment,

00:16 or taste.

00:19 But I think

00:24 aesthetics—

00:25 as well as judgment and taste—

00:27 have always been outsourced.

00:31 If someone

00:33 wants to learn about film,

00:37 what do they do?

00:39 They might search

00:40 for films others recommend,

00:46 or books people have written about film.

00:51 Through those paths,

00:55 they form their understanding of cinema.

01:01 They may also—

01:03 not may:

01:05 inevitably—

01:07 watch films made by others.

01:14 Film itself...

01:19 is an expression of taste.

01:24 It also involves aesthetics and judgment.

01:28 Directors and crews take their aesthetics

01:32 and judgments

01:34 and assemble them into the film you see.

01:38 If you watch it

01:39 and like it,

01:41 that means

01:43 the production team's

01:46 aesthetics and judgments

01:49 have been absorbed by you.

01:51 In that sense,

01:55 this too

01:56 is an example of outsourcing aesthetics and judgment.

02:00 If you don't want

02:02 your aesthetics and judgment outsourced,

02:05 make things yourself

02:08 or create works like these.

02:12 But remember:

02:14 People who can truly make a film,

02:18 a short film,

02:20 or even produce

02:22 a reasonably polished vlog

02:26 are still very few.

02:28 Everyone can shoot video now,

02:31 but not everyone can be a vlogger,

02:35 filmmaker, or short-film creator.

02:40 Likewise, everyone can take photos,

02:44 but not everyone

02:46 can be called a photographer.

02:51 Nor can everyone teach photography.

02:56 So...

02:59 Well...

03:01 To some degree,

03:02 when we appreciate something,

03:05 someone else has packaged their aesthetics and judgment

03:08 into the work we see.

03:11 And that produces

03:12 our experience of appreciation.

03:16 Returning to outsourcing:

03:20 this has always existed.

03:24 So why, in the AI era,

03:30 do we suddenly discuss it so loudly?

03:35 Perhaps AI's arrival made

03:37 some people feel threatened,

03:41 so

03:42 that created this climate of panic.

03:47 Everyone talks about aesthetics,

03:50 judgment,

03:52 and taste.

03:56 The same applies beyond film.

04:00 In other fields too.

04:02 On social media,

04:07 there are book explainers and reviewers,

04:09 reading-list curators,

04:13 and...

04:16 Simply put, intermediaries.

04:20 An intermediary is itself a form of outsourcing.

04:25 Add parentheses after it,

04:27 and many things could go inside—

04:30 not just aesthetics and judgment.

04:33 Every intermediary platform,

04:35 every middle layer,

04:37 performs an outsourcing (...) function

04:44 of some kind.

04:49 Hmm.

04:51 Earlier, I thought of

04:53 some relevant examples.

04:56 But now that I suddenly have time

05:00 to record this,

05:02 I've forgotten what they were.

05:05 Inspiration is fleeting.

05:11 I've heard this idea:

05:13 if something interesting

05:15 suddenly occurs to you,

05:17 and you don't record it immediately,

05:20 the idea may leave you

05:23 and go to someone else.

05:25 Last night, while scrolling Moments,

05:28 I saw someone repost a Moments post.

05:31 What they reposted

05:33 was an article from Dedao.

05:36 They said ideas they'd had recently

05:42 appeared almost word for word

05:44 in that article.

05:47 Maybe...

05:49 Well...

05:50 The article was certainly written

05:53 long before they had those ideas.

05:57 More abstractly,

05:59 an idea

06:01 may enter countless minds.

06:04 But

06:05 only a few people take the same idea,

06:10 think it through,

06:13 examine it critically,

06:15 find similar cases as evidence,

06:17 and write down the whole process.

06:20 So I could also say

06:25 only those who write it down

06:28 can become

06:31 the idea's

06:34 owners.

06:36 Even if everyone

06:39 thinks of the same idea,

06:42 what each person writes

06:45 will differ.

06:46 Because their way of thinking,

06:49 their logic and environment,

06:53 and their habits

06:56 are all different.

06:58 So even if everyone

07:01 writes from the same spark,

07:04 the result will differ.

07:07 As they keep writing,

07:09 each person's

07:10 direction

07:12 may change.

07:16 I feel that

07:20 if you can

07:22 catch the moment inspiration strikes

07:24 and hold onto it,

07:28 however you record it,

07:31 that's fine.

07:32 For the person recording it,

07:35 it becomes precious wealth.

07:39 I once wondered:

07:42 if someone

07:44 encounters inspiration at different moments,

07:49 how can they

07:50 capture it quickly?

07:53 Just pick up your phone,

07:58 open the recorder,

08:00 and record

08:02 whatever relates to it.

08:03 That's the fastest way.

08:07 Today we have apps

08:11 that let people dictate,

08:17 use AI to transcribe,

08:20 then turn text into structured material.

08:23 Tools like these

08:24 are among the best

08:27 for moments of inspiration.

08:33 Before vibe coding became popular,

08:36 voice input had very little visibility

08:40 or discussion in daily life.

08:45 But for some reason,

08:47 I've always felt speaking or dictating

08:51 is by far the easiest way for me to express myself.

08:55 I also have a habit:

08:58 when I don't want to type

09:01 what I'm trying to express,

09:05 I open Voice Memos and record it.

09:10 But I keep those recordings to myself.

09:12 I don't publish them.

09:16 This habit has helped me

09:19 capture many past thoughts,

09:22 emotions,

09:24 and ideas.

09:25 The problem is,

09:28 after I record them,

09:33 I never listen again.

09:37 Why?

09:39 Because listening back takes too much time.

09:42 Audio isn't something

09:44 you can skim.

09:47 If you want

09:49 to know what an episode

09:51 contains,

09:53 unless you transcribe it,

09:54 you have to listen from beginning to end.

09:57 Otherwise, you can't access its content.

10:01 Audio is...

10:05 how should I put it?

10:08 a medium with high processing costs.

10:12 It's also unusual.

10:15 I've always wondered:

10:18 why can't we have

10:20 a tool that

10:23 finds podcast content

10:27 on topics I follow,

10:28 and

10:30 returns not only relevant episodes,

10:35 but also

10:37 tells me

10:39 exactly where my point of interest

10:41 appears

10:44 in the episode?

10:48 It could even summarize

10:49 what was said.

10:52 The idea sounds simple,

10:54 but it's hard to build.

10:57 Even with AI coding so powerful,

11:02 when I build such a tool,

11:05 I still find it difficult.

11:07 One reason is:

11:09 audio, as a linear medium...

11:11 really...

11:13 to complete the whole process,

11:15 first you need an API

11:18 to interpret

11:20 what the episode

11:22 is about—

11:23 something like

11:25 transcribing it.

11:27 After transcription,

11:28 you need to use an API again

11:31 to locate

11:34 where your topic appears

11:36 and return it to you.

11:38 It may sound simple,

11:40 but the token cost is

11:45 hmm...

11:46 substantial.

11:47 Previously,

11:50 while testing a tool's MVP,

11:55 I only asked it to analyze—

11:57 to process

11:59 one episode

12:00 under an hour.

12:02 When I checked the bill,

12:04 several dollars were gone.

12:06 Testing one sub-hour episode

12:10 cost several dollars.

12:12 That's expensive.

12:15 So I wonder if

12:17 there's a faster way

12:19 to bring this ideal tool into existence.

12:24 I don't know.

12:28 Ha, this episode really...

12:32 I can't remember

12:33 what I said at the beginning.

12:35 Anyway...

12:37 how does that saying go?

12:40 I just ramble wherever my mind goes.

12:43 All right.

12:46 I'll continue next time.