As AI-generated songs flood streaming platforms, artistes defend the human touch that algorithms cannot replicate
Press play. Music fills the room. The vocals sound convincing, the arrangement is polished, the production clean and balanced. It carries the familiar gloss of a recording studio release.
Press play again. Another track follows, equally assured, equally well-produced. One question lingers: which was created by a human and which was generated by artificial intelligence? For most listeners, the answer is increasingly unclear.
Music streaming platform Deezer reported in April 2026 that around 75,000 fully AI-generated tracks were being uploaded daily, making up about 44% of all new uploads on its platform. In a separate blind listening survey, 97% of respondents struggled to distinguish between AI-generated music and songs created by humans.
Artificial intelligence can now compose melodies, generate lyrics, imitate voices and produce entire songs within seconds.
Platforms such as Suno and Udio allow users to create music from simple prompts, while AI-assisted tools are widely used for mastering, transcription, arrangement and songwriting.
Yet despite the speed and sophistication, musicians and educators insist that something essential remains missing.
AI improves creativity but does not replace humans
Dr James Boyle, dean of the music faculty at Aswara, views AI as part of music’s long technological evolution.
From analogue recording and synthesizers to digital audio workstations, music has always shifted alongside technology.
“It always has,” he said when asked whether technology had improved creativity.
But Boyle, son of legendary Malaysian jazz musician Jimmy Boyle, is clear that AI should remain a supporting tool rather than a substitute.
“As for creativity, AI is good as a tool for sketches and ideas of exploring new possibilities, but it is not an end product. Human talent takes precedence,” he said.
He describes AI-generated music as still “formulaic and mechanical.”
Sharing a similar view, Universiti Sains Malaysia music department chairperson and lecturer Tan Jin Yin said musicianship depends on human judgement that machines cannot replicate.
“A musician still needs to decide what is meaningful, what is stylistically suitable, and what communicates honestly,” she said.
While acknowledging that AI-generated music can sound technically convincing, she said emotional depth is harder to fake.
“I have listened to AI-generated songs with very skilful vocals and polished production but they still seem to lack a certain genuine emotional depth.
Technically, the sound may be convincing but emotionally it can feel slightly empty or detached as though the music understands the shape of emotion without truly experiencing it,” she said.
The algorithm versus the organic
For Jerome Quah, president of the Penang Jazz Society, the difference lies in unpredictability.
“Humans are unpredictable. When it comes to creating music, everyone is unique in their own way, whereas AI is based on algorithms. AI creates from what already exists. Human creation can be something totally new,” he said.
He stresses what he calls the “organic-ness” of music-making.
While AI is mechanical and predictable, he said human creation is “very organic and original”.
When asked whether AI could replicate emotion, lived experience and interpretation, his answer is direct: “No.”
Quah believes music is rooted in instinct, subconscious response and emotional spontaneity—qualities that cannot be fully encoded.
“The AI is just a machine. It can never be like humans as it lacks subconsciousness, and it doesn’t have a soul,” he said.
Jazz as the last stand
For Quah, jazz remains one of the most difficult genres for AI to genuinely replicate.
Founded in 2016, the Penang Jazz Society has around 50 members and runs weekly practice sessions, jazz training ensembles and bi-monthly jam sessions.
Its Big Band of Penang has performed at regional festivals including the Thailand International Jazz Conference and Hua Hin International Jazz Festival.
Jazz, he said, depends heavily on improvisation and shared instinct in real time.
“It is a spontaneous art form. Musicians rely on the combination of musicians and the inspiration at the moment. Maybe one day, AI will be able to replicate that but replicating still means relying on algorithms it already knows,” he said.
During live performances, jazz musicians respond to one another instinctively, shaping music in the moment. That interaction, he argues, is still beyond artificial systems.
“So it will never be able to play jazz successfully,” he said.
However, he concedes that AI may eventually assist in arranging jazz compositions for larger ensembles.
The problem with AI in education
If AI struggles with emotion in performance, it raises different concerns in education.
At USM, some students already use AI tools for music production, notation, composition and research, though others still prefer manual composition and live performance.
Tan allows AI use, provided students are transparent.
“To me, the important issue is not simply whether students use AI, but whether they understand what they are doing,” she said.
She warns that AI-generated responses may appear fluent but often mask a lack of understanding.
“In music theory, students cannot only give a beautiful explanation.
“They need to understand why a chord progression works, how a phrase is shaped, or how certain musical elements function in context,“ she said.
To test genuine understanding, she probes students beyond their submissions.
“If the student can explain the answer clearly in their own words, apply the idea to another example, or justify their musical choices, then I know learning has taken place.
“But if the student cannot answer basic questions about their own submission, then it usually shows that the work was copied and pasted without real understanding,” she said.
She emphasises that AI must not replace the learning process itself.
“In music education, the process of listening, analysing, questioning, and making musical decisions is still very important,” she said.
The shortcut generation
While all three agree AI has useful applications, they also warn of growing dependency.
Boyle cautions that convenience can quietly erode creativity.
“When there is technology, the temptation to utilise it, or to some, even to abuse it, is plentiful and highly contagious. To a few others, it could hinder their real creativity because of the unprecedented shortcuts being offered by AI,” he said.
Quah adds that younger musicians may lose exposure to traditional, manual processes.
“The young generation who use all these tools have never experienced laborious manual work as the older generation did.
“They will never understand the working of composing, which was entirely manual,” he said.
Still, none reject AI outright. Instead, they argue for balance.
Boyle offers a final reflection: “As long as we don’t feed the machine everything, we still have the upper hand.”
For now, at least, musicians appear determined to keep it that way.
