What was the first AI-generated song? The real history of computers making music

Before AI music: computers learned to play sound

Computer music began before modern artificial intelligence. The Computer History Museum notes that computers were already being used to perform music in the early 1950s; a 1951 performance of Colonel Bogey March by Australia’s CSIRAC was among the earliest examples. That was an important technological step, but playing or synthesizing a tune is not the same as composing one.

1956–1957: the Illiac Suite and the first computer-composed score

At the University of Illinois, Lejaren Hiller and Leonard Isaacson used the ILLIAC I computer to generate musical material under programmed rules. The result was the Illiac Suite for string quartet. University of Illinois sources describe it as the first computer-composed piece or first score created with a computer.

The dates are sometimes given as 1956 and sometimes 1957 because the work was created and first performed in 1956, while its methods were formally presented and published around 1957. It belongs to the history of algorithmic composition rather than the modern text-to-music systems people now associate with generative AI.

1961: Daisy Bell was the first famous computer voice, not an AI composition

Daisy Bell is often pulled into the “first AI song” story because Bell Labs programmed an IBM computer to sing it in 1961. It became an iconic demonstration of computer speech and music synthesis and later inspired the HAL 9000 scene in 2001: A Space Odyssey.

But Daisy Bell itself was written by Harry Dacre in 1892. The computer performed an existing song. It did not compose the song, so calling Daisy Bell the first AI-generated song is inaccurate.

2011: Iamus and a work composed entirely by a computer

A much stronger answer to the “completely composed by a machine” question arrived with Iamus, a system developed at the University of Málaga using Melomics technology. Its full-scale piece Hello World! was composed in 2011 and premiered that year.

Research associated with the project describes Hello World! as arguably the first full-scale work entirely composed by a computer and automatically written as a conventional score. Human musicians performed the result, but the compositional material itself was generated by the system. The University of Málaga later presented the Iamus album as the first record made entirely of pieces composed by a computer.

2016: why Daddy’s Car became the “first AI pop song”

Daddy’s Car is the milestone most likely to appear when someone searches for the first AI song. Sony Computer Science Laboratories presented it in 2016 as a pop song composed with its Flow Machines technology, which learned from a database of roughly 13,000 lead sheets.

This is where the word “completely” matters. Sony CSL’s own account says composer Benoît Carré selected the musical style, FlowComposer generated a lead sheet with melody and harmony, a second system helped match audio material, and the human musician then finished the production and mixing. Carré also wrote the lyrics. Daddy’s Car was therefore a major AI-assisted composition milestone, but not an end-to-end song created without human creative decisions.

• If you mean “first widely publicized AI-composed pop song,” Daddy’s Car is a defensible answer.
• If you mean “first complete work composed autonomously by a computer,” Iamus and Hello World! are earlier.
• If you mean “first score composed with a computer,” the Illiac Suite goes back to the 1950s.

2023: Heart on My Sleeve made AI music a mass-culture story

In 2023, Heart on My Sleeve became one of the first AI-music tracks to go truly viral at global scale because it used synthetic vocals resembling Drake and The Weeknd. Guinness World Records documented tens of millions of views across platforms around the viral moment.

It was not the first AI song, and it was not a fully autonomous machine creation. Its importance was cultural: millions of listeners suddenly confronted a song in which artificial voices could sound like famous performers without those performers taking part.

So what was the first famous song made completely with AI?

The most accurate answer is that there is no universally accepted single title, because “completely made with AI” changes meaning across eras. Older systems often generated notation that humans performed; later systems generated parts of composition; modern generators can synthesize the final audio itself.

If a reader asks for one famous early pop milestone, Daddy’s Car is the clearest reference — with the essential warning that humans still made major creative and production decisions. If the question is about autonomous computer composition, Hello World! by Iamus is a stronger earlier candidate. And if the question is simply where computer composition began, the Illiac Suite is the foundational landmark.

Why this history matters now

Today a creator can describe a song in natural language and receive finished audio in minutes. That makes the technological jump feel sudden, but the underlying story is almost seventy years old: rules, probability, synthesis, machine learning, style modelling and finally generative systems that combine many of those ideas.

Knowing the milestones also helps avoid a common mistake: “AI music” is not one single technique. A track can use AI for composition, voice conversion, separation, mastering, lyrics, arrangement or complete audio generation. When judging authorship, rights or originality, those differences matter.