We Only See What’s on Stage, Not What Happens Backstage
Recorded in Chinese
AI-sounding posts lead me to reflect on judgment, taste, and the editing behind podcasts. I also revisit a seaside piano that became a personal memory, and why I want an outlet for my own voice.
00:00 I've always wondered:
00:04 when people scroll through text
00:06 on social platforms,
00:09 why do posts
00:12 that feel heavily AI-generated
00:16 still receive
00:17 so many likes and saves?
00:20 I genuinely don't understand.
00:25 In my own experience,
00:27 if I come across such posts,
00:30 I would never like or save them,
00:33 no matter how informative they are.
00:36 I feel that
00:37 if you're seriously making the content,
00:41 you should take what you want to share,
00:44 put it in your own words, then post it.
00:49 I truly cannot understand
00:52 why something that feels so AI-generated
00:57 can still receive
00:58 so much positive feedback.
01:03 Another thing is,
01:05 when people hire today,
01:11 most roles
01:13 care whether someone has AI experience.
01:18 But I just saw another post.
01:23 The writer had attended a talk
01:28 by AMD's CEO at a university.
01:32 I believe her name is Lisa Su.
01:38 I don't know much about her.
01:40 But according to the post,
01:42 the CEO said
01:45 one ability she values in hiring
01:49 is whether someone can tell
01:54 if an AI-generated result is correct.
01:56 I don't know her exact words,
01:59 because the person sharing them
02:03 used AI to summarize the talk.
02:06 So my rough interpretation is:
02:10 can someone determine
02:12 whether an AI's output is
02:14 right or not right?
02:17 If that person
02:19 can tell,
02:20 then they clearly possess judgment.
02:23 Their judgment is fully active.
02:26 I feel the same way.
02:28 If someone chooses to create content
02:32 and uses AI to assist them,
02:37 if they can guide the AI
02:40 so that the resulting writing
02:43 doesn't feel AI-generated,
02:46 only then have they truly used AI well.
02:50 But if they can't even judge
02:54 whether the AI's output feels AI-made—
02:57 whether it carries that AI flavor—
02:59 then I think
03:02 their way of using it
03:03 has completely failed.
03:06 Truly.
03:09 So much writing on Xiaohongshu
03:12 feels intensely AI-generated.
03:16 Especially when
03:17 the discovery feed
03:20 shows several posts in a row
03:23 written or assisted by AI,
03:25 I think, my goodness,
03:27 what is going on with these people?
03:31 Even if AI generated it,
03:33 they could at least revise it.
03:35 But they didn't.
03:37 That really frustrates me.
03:41 Especially when writing like this
03:44 still receives so much positive feedback,
03:47 I feel the world
03:49 has turned upside down beyond imagining.
03:55 I remember someone on WeChat Moments
04:00 who had been an editor at a major tech publication.
04:05 They once criticized this phenomenon.
04:07 They said Xiaohongshu had
04:09 so much AI-assisted
04:14 or directly AI-written content,
04:17 and after receiving that text,
04:19 people didn't even revise it.
04:23 They felt deeply disappointed,
04:27 and even a little angry.
04:31 Personally, I find it
04:34 genuinely, deeply frustrating.
04:37 So whenever I encounter someone
04:39 who truly puts care into their writing,
04:42 I find that person incredibly precious.
04:45 Truly precious.
04:47 And sometimes I see
04:49 their numbers fall far behind
04:53 those highly AI-flavored posts.
04:55 And I think, my goodness,
04:56 what a shame.
04:58 This world—
05:01 I'm tempted to use that metaphor again—
05:04 has turned upside down beyond imagining.
05:09 Can the people liking and saving them
05:13 really
05:14 not detect the AI flavor?
05:19 Or have they simply
05:22 accepted that such things exist?
05:29 I honestly don't know.
05:33 Why have they accepted it?
05:39 I can't understand.
05:44 Returning to the CEO's point,
05:46 I think it's truly important.
05:49 Whether something a person creates
05:53 still feels human
05:55 can actually reveal
05:58 something about the person,
06:02 far beyond their judgment.
06:05 It shows their aesthetics
06:06 and their taste.
06:08 These things are immensely important.
06:13 This includes some public accounts,
06:21 and even content on X.
06:30 How should I put it?
06:33 I used to wonder why AI
06:36 so often says "not A but B,"
06:39 or uses an "A but not B" structure.
06:44 I haven't spent a long time
06:47 immersed in an English environment,
06:50 so I don't know how often
06:52 such phrasing occurs in daily life.
06:56 But at least when I read
07:00 English writing on Substack,
07:02 I rarely see anything similar.
07:04 Perhaps on X,
07:07 a social platform built around short text,
07:10 that kind of structure may appear.
07:13 But my own feeling is
07:15 that it is still fairly uncommon.
07:19 Yet AI often produces such sentences.
07:23 I don't know whether those sentences feel strongly AI-generated.
07:29 But at least when some AIs
07:32 interact with us
07:35 in Chinese,
07:36 the text they provide
07:40 was actually translated from English
07:42 and then presented to us in Chinese.
07:45 During that process,
07:46 it inevitably carries traces of English,
07:50 or an English shadow.
07:52 Chinese usage and word order
07:54 differ from English in some ways—
07:57 subject, verb, object, for example,
07:59 and various other structures.
08:03 You can often tell very clearly
08:07 that they are different.
08:09 Sometimes, reading an entire piece,
08:12 your instincts say something is wrong.
08:15 It feels wrong because the result
08:19 simply doesn't sound Chinese.
08:27 I don't know.
08:30 Maybe everyone accepts this world now,
08:34 or perhaps this is simply their taste.
08:44 Sometimes I even wonder—
08:47 because
08:50 I don't want to
08:53 reveal what kind of work I do.
08:58 Oh.
09:00 All right.
09:01 My work is actually related to audio.
09:04 I listen to a lot of podcasts,
09:08 and I may also
09:09 be involved in podcast production.
09:12 So I get to see
09:14 what hosts' and guests'
09:18 raw, unedited speech is like.
09:21 Actually,
09:22 how should I put it?
09:23 Before doing this work,
09:26 I always assumed the podcasts I heard,
09:30 the way
09:32 hosts spoke
09:33 and guests spoke,
09:34 were exactly as they sounded.
09:36 But after I actually began
09:38 working in this area,
09:39 I realized—
09:41 wow.
09:42 I was stunned.
09:45 Sometimes a very articulate person
09:48 appears before you
09:52 only after its image is filtered.
09:55 Whether in podcasts
09:56 or the videos we watch online,
09:58 or the words we read,
10:01 the same is true.
10:03 The form
10:05 in which someone appears before you
10:09 has been carefully selected
10:11 and designed by them.
10:12 We see only the front stage,
10:16 not everything behind it.
10:19 I mean,
10:22 people often assume
10:25 that what they see is the whole person,
10:28 their true level of ability.
10:30 We fail to imagine
10:34 what exists backstage
10:39 behind the public image.
10:43 Now, instead,
10:45 with anything I encounter,
10:46 unless I have
10:47 truly seen
10:51 as much as possible,
10:52 I won't completely believe it.
10:55 There will always be doubt,
10:58 or a reluctance to believe it completely.
11:01 Because once you
11:04 have encountered enough things—
11:06 take podcasts, for example—
11:10 how exactly
11:11 how does an episode get made?
11:13 After experiencing the full process,
11:15 the finished podcast
11:18 becomes demystified for me.
11:23 I still trust
11:25 what the podcast says,
11:28 its quality,
11:29 its expression,
11:30 and its presentation.
11:32 But I won't
11:34 idealize the two people I hear,
11:38 or however many people appear.
11:41 Because I believe—
11:43 or rather, because I know—
11:45 that some design
11:49 and some selection happen behind it
11:51 before we are shown the result.
11:57 Social media is the same.
11:59 People have argued for years that
12:04 some users say
12:05 what a person chooses to post
12:07 on social media
12:09 is a performance,
12:10 or an attempt to gain attention.
12:14 In some ways, that's true,
12:19 but in other ways, it isn't.
12:22 Some people genuinely,
12:23 from the heart,
12:25 want to share the beauty they observe
12:29 and the beauty they experience.
12:31 They simply place it there.
12:33 It may be nothing more than a record.
12:36 Such people absolutely exist.
12:44 Recently, for example,
12:46 I've been planning an article
12:50 and consulting
12:52 or exploring some sources.
12:54 While reading them,
12:57 I came across—
13:00 mm, I forgot the scholar's name.
13:05 Anyway, they wrote several books.
13:07 One of their central arguments
13:12 was that the medium is the message.
13:17 Take what we just discussed,
13:22 uh—
13:24 what did I mention at the beginning?
13:28 I can't remember.
13:30 All right.
13:31 Take podcasts, for example.
13:34 The sound we hear
13:37 is a medium carrying podcast information.
13:41 The sound itself is information,
13:45 and its design and arrangement are too.
13:50 They simply aren't
13:54 what we conventionally call information.
13:58 Whenever something exists,
14:03 it is both a medium and a message.
14:07 Let me give an example.
14:11 While traveling,
14:15 I took a walk by the sea.
14:17 At one spot along the shore, I saw
14:22 a broken piano.
14:27 In front of it
14:28 was an empty chair.
14:31 The chair and piano were arranged so that
14:37 if someone sat in the chair,
14:39 they would face the sea.
14:42 When I saw that scene,
14:44 it sparked some thoughts.
14:46 For example,
14:46 isn't there a film
14:48 called The Legend of 1900?
14:50 So I wondered,
14:54 "Has the pianist gone out to sea?"
14:56 The scene brought to mind
15:00 memories already associated with it
15:02 and led me to other meanings.
15:06 That meaning was something I created:
15:08 "Has the pianist gone out to sea?"
15:10 It was a slightly playful interpretation.
15:15 First, I saw the scene,
15:17 and then I created meaning.
15:19 My impression of the city
15:21 became much deeper, right?
15:24 Even long after leaving the city,
15:28 I could remember
15:29 walking by the sea one day,
15:32 seeing that scene,
15:33 and generating
15:34 that meaning from the image.
15:37 The city's image
15:40 would remain in my mind for years.
15:43 What does that show?
15:45 The city's administrators,
15:49 or managers,
15:51 or the administrators of that district,
15:54 had achieved their goal.
15:56 Or perhaps
15:57 it came from somewhere nearby,
16:00 such as a building owner.
16:03 I don't know who placed
16:06 the piano and chair there.
16:07 I imagined many possibilities.
16:09 Perhaps it was the city
16:12 or district management.
16:14 Or perhaps a nearby household
16:18 had a broken piano
16:20 perhaps,
16:21 moved it there.
16:22 Its original value was already gone.
16:25 It wasn't an antique collectible.
16:28 So placing it there
16:30 made it part of this tourist city's—
16:35 mm,
16:37 what was the term?
16:40 A cultural landmark.
16:44 That design became
16:46 a cultural landmark for the city.
16:48 You could say that.
16:50 If such a scene
16:53 remains in my mind for years,
16:56 even if I haven't
16:58 visited the city for a long time,
17:00 I will still remember it.
17:02 That means the goal was achieved.
17:05 It allowed the city
17:07 to take root in some travelers' minds.
17:13 Such an image,
17:15 and such objects—
17:16 these concrete things—
17:17 are also a medium.
17:20 And the medium itself is information.
17:24 That information is the meaning I create.
17:28 I mean,
17:33 I've always felt
17:34 that every concrete object
17:36 you encounter
17:38 contains points of information.
17:40 For example, a person you know
17:43 also contains points of information.
17:45 How can that information be interpreted?
17:47 Don't we often say
17:49 that we never meet someone
17:52 for no reason?
17:54 That person's presence
17:55 in our life has a reason.
17:56 And that reason is information.
17:59 That reason
18:02 is attached to the person
18:04 and interpreted through them.
18:07 So the person is also information itself.
18:13 Oh, I've forgotten
18:15 what my point was.
18:16 What was that thing again?
18:18 I keep trying to bring the topic
18:19 back to what I said at the beginning,
18:23 but I no longer remember
18:24 what I said at the beginning.
18:29 What was I talking about?
18:34 Xiaohongshu posts.
18:37 Oh, posts on that app.
18:40 Oh, right.
18:41 I began by talking about
18:44 judgment and aesthetics in the AI era.
18:49 But how did I end up
18:51 at "the medium is the message"?
18:57 How should I
18:58 connect these two ideas?
19:06 Okay, I'll force a connection.
19:11 When we
19:14 scroll through social media,
19:15 we see pieces of written information
19:17 that other people put out.
19:20 By reading
19:21 and interpreting that text,
19:24 you can see
19:26 what kind of person
19:28 posted it.
19:31 And what kind of person they are
19:33 is simply another set of information:
19:37 their aesthetics,
19:40 their judgment,
19:41 and their taste.
19:45 So the text we see
19:48 isn't merely the text
19:50 someone chooses to display.
19:53 Behind it, the text represents the person.
19:58 The person,
20:00 what they choose to show,
20:02 and the thought they put into
20:04 arranging the words—
20:07 or the choices they weigh—
20:09 all of these are information.
20:17 Whether we're talking about
20:18 a specific person,
20:20 the platform carrying the content,
20:23 or the meaning we create
20:26 through what we see,
20:32 each is both a medium and a message.
20:38 I no longer know how to tie this together.
20:43 Let's stop there.
20:44 I'll just say whatever comes to mind.
20:46 I once tried
20:48 listening to Jiang Sida's podcast.
20:52 I was shocked.
20:55 Why did so many people listen to it?
21:00 He truly is someone
21:02 who says whatever comes up
21:04 the moment it comes to him.
21:06 I even saw comments
21:08 under the paid podcast
21:11 where many people said
21:12 he spoke without logic,
21:14 ideas,
21:15 or any clear train of thought.
21:17 I mean—
21:18 well, he says whatever comes to mind.
21:23 Yet they were still willing to pay.
21:26 I couldn't understand it.
21:28 How could that be?
21:31 I know very well that Jiang Sida
21:32 does have his own personal charisma.
21:36 But where does that charisma appear?
21:41 I don't know either.
21:44 I truly don't know.
21:45 But if one day
21:47 someone genuinely feels
21:48 that something I say inspires them,
21:51 I would be incredibly happy.
21:57 I've always felt that speaking
22:01 is more comfortable than writing for me.
22:05 With writing,
22:07 how should I put it?
22:10 Before writing every sentence
22:12 and every word, I think very deeply.
22:16 I consider its logic
22:18 and whether the sequence feels wrong.
22:27 I hold writing to very high standards.
22:32 So every time I write an article,
22:34 I think about it a great deal.
22:36 Even if
22:37 I've planned an article for a month,
22:41 I still may not have written it.
22:43 But when I express myself through sound,
22:45 I can, in a short time,
22:47 express an enormous amount.
22:50 Just like
22:52 that classic moment just now,
22:55 I had no idea what
22:57 I said at the beginning.
22:58 But once I start talking,
23:01 I can
23:05 keep talking nonstop.
23:06 I simply cannot stop.
23:08 Unless I choose to wrap up,
23:10 I can keep talking forever.
23:14 When I was young,
23:16 and my grandmother was still alive,
23:18 she would ask me,
23:20 "Why do you sometimes talk to yourself?"
23:25 I was a little surprised.
23:27 Did I really talk to myself?
23:31 I knew I sometimes did it deliberately.
23:35 But other people had heard me
23:39 talking to myself
23:40 and told me about it,
23:42 yet I had no memory of those moments.
23:44 Sometimes I simply
23:46 don't know I'm talking to myself.
23:49 But at other times,
23:50 I'm fully aware of doing it.
23:53 Talking to myself
23:54 can be something I initiate,
23:57 but I didn't realize
23:59 it could also begin unconsciously.
24:02 I don't know.
24:03 Anyway, speaking aloud
24:05 is an extremely comfortable
24:07 way I express myself.
24:10 Sometimes I even think,
24:11 if someone
24:13 wanted to make a podcast
24:16 but
24:17 couldn't find the right person—
24:19 for example, someone to host it—
24:22 I would be more than happy
24:24 to take on that role.
24:26 And I can promise
24:27 I would make the episode excellent.
24:32 I'm especially good
24:37 at presenting someone in my own way—
24:41 taking the person I see
24:44 and want to understand,
24:46 and presenting them fully to listeners.
24:52 I wonder when such a dream job
24:55 will ever come to me.
24:58 Of course, I could also
24:59 start such a project myself.
25:04 But there are
25:05 so few people like that around me.
25:08 I mean,
25:09 I have my own standards for choosing people.
25:16 How should I put it?
25:18 Those standards are extremely strict.
25:20 Sometimes I want to scold myself.
25:24 Why are my standards so high?
25:32 So now I plan
25:36 to release all this rambling.
25:39 It's also an attempt to overcome
25:41 my own high standards.
25:46 I mean,
25:51 behind what I choose to share,
25:55 there is so much more.
25:58 Some of it might be valuable if shared,
26:02 but I think it isn't perfect enough,
26:05 informative enough,
26:08 or valuable enough,
26:09 so I hide it away.
26:12 But that isn't a good thing for me,
26:17 because
26:22 mm,
26:25 I need to build ways to be seen.
26:29 I don't know whether this
26:33 Vaiha project will be noticed,
26:36 but doing it now
26:40 makes me happy in itself.
26:44 It is also a way
26:45 to create an outlet for myself,
26:49 express what I want to express,
26:52 and say what I want to say.
26:54 I think that alone
26:58 is already meaningful enough.