Programmers Treat AI as a Power Tool; Musicians See It as an Impersonator

OK, thank you all for your long-term support and attention 🌹. Recently, I came across a very typical debate...

Someone said: "Programmers can accept AI writing code, so why can't musicians accept AI writing songs? Are they being too pretentious?"

This statement sounds reasonable at first glance.

After all, programmers complain that AI-written code is "unreliable" while integrating Codex, Copilot, Cursor, Claude Code, and Tongyi Lingma into their workflows. Musicians, on the other hand, acknowledge that AI melodies, arrangements, and voice mimicry are getting better and better, yet they keep emphasizing that AI music must be labeled, must be authorized, and should preferably be placed in a separate "AI music pool" rather than mixed into normal playlists.

Both involve generative AI, so why does it feel like a "new power drill" in the coding world but like "a stranger barging into the rehearsal room" in the music world?

I want to start with a judgment that might be controversial:

Musicians resist AI not because they don't understand technology; often, it's because they see the cost earlier than bystanders.

This isn't about who is more conservative or more progressive.

The fundamental reason is: Programmers and musicians are not facing the same kind of AI.

1. Programmers embrace AI because code inherently has a "verification mechanism"

When AI writes a piece of code, a programmer's first reaction is usually not "it stole my soul," but: Can it run? Are there bugs? Are there security risks? Do the tests pass? Will it crash in production?

The coding world has a very cold verification mechanism: compilers, unit tests, integration tests, logs, monitoring, code reviews, and rollback mechanisms. No matter how beautifully AI generates code, if it doesn't run, it's garbage.

This is also the daily reality for many programmers: meetings during the day, changing requirements at night, staring blankly at error messages in the early morning. At times like these, AI can help explain an unfamiliar framework, complete a block of repetitive code, or translate a mess of configurations into plain language. It naturally becomes addictive.

So, a programmer's feeling towards AI is more like: It's a very fast intern with a great memory but who is often confused.

You can let it write boilerplate code, explain errors, supplement tests, look up documentation, and make refactoring suggestions, but the responsibility for whether to merge the code ultimately rests with the person. If a company has an incident, they won't ask AI to attend the post-mortem meeting; the developers, testers, architects, and managers will bear the responsibility.

This is why programmers say "AI will replace me sooner or later" but honestly use AI in their work. Because most of the time, AI isn't directly taking away their authorship; it's helping them save some repetitive labor within a process where they already bear the responsibility.

This point is very crucial.

What programmers worry about is: Will I be replaced by someone who is better at using AI?

What musicians worry about is: Will my voice, my style, and my training traces become fuel for others to mass-produce low-cost content without my knowledge?

These two fears are not on the same scale.

2. Musicians resist AI not just out of "fear of unemployment," but fear of their work being diluted...

Many people underestimate the pain musicians feel.

Someone who writes code might say: "I also spent ten years learning to program, why can I accept AI?"

But if you change the question, it feels different:

If one day, a "singer" with a voice almost identical to yours appears online, singing new songs with your vocal style, taking advertisements with your style, and gaining followers with your shadow, but you haven't authorized it and receive no profit, would you think that's just technological progress?

But the commercial logic of music and code is different.

The value of code usually comes from functionality, stability, delivery, and maintenance. Users won't pay because a function "looks like it was written by John." Most business code doesn't rely on personal style being recognized by the public.

Music is exactly the opposite.

The value of a song lies not only in the melody and lyrics but also in the voice, vocal technique, articulation, arrangement habits, era memories, and even the fans' projection onto the person. Music sells not just an "audio file" but a connection between people.

The most sensitive point about AI music is right here: It doesn't just generate "things that sound like music"; it can also generate "music that sounds like a specific person."

If a platform suddenly has a large number of "songs like Jay Chou," "voices like Eason Chan," or "arrangements like a certain indie musician," even without direct attribution, it's enough to squeeze the original artist's recognizability. For a musician, this isn't ordinary competition; it's their creative traces being copied, diluted, and repackaged.

But... many ordinary listeners don't necessarily care who is behind it. They just say: "It sounds pretty good."

For musicians, the most heartbreaking thing might not be that AI can write songs, but that the voice and aesthetic they spent over a decade cultivating are ultimately compressed into a single prompt: "Give me a song in a certain style."

Programmers often see AI in terms of efficiency.

Musicians first see AI in terms of boundaries.

3. The copyright issue hasn't been clarified yet, so of course the music industry dares not let it pass freely!

Over the past two years, the reason AI music controversies have repeatedly appeared on the edge of trending topics is not because musicians are pretentious, but because the industry really hasn't sorted out the accounts.

In 2024, major US record labels filed lawsuits against AI music companies like Suno and Udio, with the core dispute being whether the training data used copyrighted sound recordings. The reason the entire industry is watching this case is that it relates to a fundamental question:

Before AI generates a song, whose works has it consumed?

If it has consumed them, was it authorized? If not, how are the profits distributed? If the generated result is highly similar, who bears the responsibility? If it's just "stylistically similar," how does the law judge that?

This set of problems also exists in the coding industry, such as open-source licenses, training data, code similarity, and commercial licensing. But the coding industry at least has a large number of traceable repositories, protocols, and dependency management tools, and enterprises can reduce risks through private models, code scanning, and compliance audits.

The music industry is more troublesome.

Because similarity in music isn't just a problem when it's "copy-paste." A few notes, a chord progression, a vocal style, or a segment of timbre can all cause disputes. Not to mention voice cloning, which directly touches on personality rights, portrait rights, performer's rights, and fan trust.

This is why when Universal Music Group and TikTok re-reached an agreement in 2024, they publicly emphasized protecting human artistry, removing unauthorized AI-generated music, and improving attribution for artists and songwriters. Translated into plain language, it means: It's not that AI isn't allowed in, but you must first clarify who authorizes, who benefits, and who is responsible.

Musicians demanding an "AI music pool" is essentially not about locking the technology away, but about putting a fence around content for which the rules haven't been set yet.

4. Programmers use AI as "I command the tool"; musicians fear AI because "the tool impersonates me"

This sentence might be a bit harsh, but it's very close to the real difference.

When a programmer uses AI to write code, it's usually done within their own project, their own IDE, and their own requirement context. What AI generates is an intermediate product. It must be reviewed, tested, and modified by a person before it can become the final deliverable.

When an AI music model generates a song, the situation is different. What the listener hears is often already the final work. Many people won't care if there's authorization behind it, won't know who it imitated, and won't actively distinguish "was this sung by a person or a model?"

This creates a strong sense of insecurity for musicians:

I haven't even stepped on stage yet, but another voice "like me" has already started performing.

And it doesn't get tired, doesn't have mood swings, doesn't negotiate splits, doesn't need rehearsals, doesn't need tours, and won't get a hoarse throat. It can generate a thousand songs a day, filling short videos, live streams, mini-games, ad background music, and mood playlists at a low price.

For platforms, this is certainly tempting.

For musicians, this is danger.

Because when content supply approaches zero cost, what's truly scarce is not "whether there are songs" but "whether people are still worth being seen."

5. Why has "isolating an AI music pool" become a realistic choice?

Many people hear "isolation" and think it's conservative, as if the music industry is practicing technological discrimination.

But from a platform governance perspective, an AI music pool actually has practical value.

It at least solves one fundamental problem: Don't let users unknowingly treat "humans" and "models" as the same type of content consumption.

First, we need to know that it protects the user's right to know. You can like AI songs, but you should at least know whether what you're hearing is AI-generated, AI-assisted, or human-created.

Second, it protects the market position of creators. If AI content and real human works are completely mixed together, platform algorithms will naturally favor content that is large in quantity, low in cost, and fast to update, squeezing the survival space for newcomers in the long run.

Third, it facilitates copyright and revenue settlement. Whether AI works can be commercialized, enter playlists, participate in revenue sharing, or run advertisements all require separate rules. Although music apps like Kugou, NetEase Cloud, and Qishui have already seen a large number of AI songs, they are struggling to generate benefits. AI is still AI...

Fourth, it buys the industry negotiation time. It's not that AI music can't develop, but training data, licensing models, voice cloning, attribution mechanisms, and revenue split ratios cannot rely on a "pollute first, clean up later" approach.

So, the AI music pool is not the endpoint; it's more like a fire door for the transition period.

This door is not meant to be closed forever, but to prevent the entire building from catching fire before everyone has agreed on the rules.

6. The real divergence is not "whether to use AI" but "who defines how AI is used"

I know some music practitioners who actually don't reject AI.

Some use AI to find melodic inspiration, some use it to make demos, some use it to clean vocals, separate audio tracks, assist with mixing, and some use AI to quickly try different arrangement directions.

What they resist is not the "tool," but the "unauthorized substitute."

This statement actually applies to many ordinary people as well. We don't dislike a tool that saves us effort, but we find it hard to accept something else taking our face, voice, experience, and identity and then telling us: This is the trend of the times, you need to adapt.

This isn't contradictory to programmers. Programmers also don't like companies using AI to monitor performance, don't like their code being used for training opaquely, and don't like bosses thinking "with AI, we can hire half as many people." Programmers embrace AI as controllable productivity; musicians resist AI as an uncontrollable replacement relationship.

So, this matter cannot be simply summarized as: Programmers are more open, musicians are more conservative.

A more accurate statement is:

Programmers' AI is mostly still on the workbench. Musicians' AI has already stepped onto the center of the stage.

One helps you work. The other might pretend to be you to be liked.

This is the source of the emotional difference.

7. My judgment on AI music: It will explode, but it won't integrate painlessly.

AI music will definitely continue to develop and will become stronger and stronger.

In the future, there will be a large amount of AI-native music and digital singers: soundtracks generated in real-time for games, mood songs automatically matched for short videos, personalized sleep music, fitness music, and companion audio content. In many scenarios, people don't need a "star," just a piece of sound that fits perfectly.

But at the same time, human music won't disappear because of this.

The more AI can generate infinitely, the more it will contrast the value of live human performances, real experiences, unique personalities, and long-term companionship. In the future, listening to music might split into two needs: one is functional sound consumption, the other is relational emotional connection.

AI will be very strong in the former. For the latter, people remain important.

It's just that the music industry must first establish the rules:

Training requires authorization, voice mimicry requires consent, generation requires labeling, distribution requires partitioning, revenue must be traceable, and infringement must be accountable.

Without these prerequisites, so-called "embracing AI" is not an embrace for many musicians, but being pushed into the water.

Hmm...

Programmers embrace AI because most of them can still fit AI into their own processes.

Musicians resist AI because they are afraid of being fitted into AI's process.

The difference here is not technical understanding, but security, profit rights, and dignity.

The best future for AI shouldn't be making all industries surrender in the same posture. It should allow different professions, within their own boundaries, to choose how to cooperate, how to label, how to split revenue, and how to preserve the human position.

So, I support the development of AI music, and I also support the AI music pool.

Not to exclude new technology, but to make new technology learn to respect those already on the field before it enters.

Finally, I'll leave three questions, and everyone's answers might be different.

First, can you accept AI singers entering your daily playlist?

Second, if a song sounds great but isn't sung by a human, would you mind?

Third, if AI imitates a singer you've liked for over a decade, would you still think "it sounds good, so it's fine"?

My own answer is: I don't oppose AI writing songs, but I hope it stands in its own position and doesn't enter wearing someone else's shadow.

Do you support AI music being separately labeled and partitioned? Feel free to share your honest thoughts in the comments section.