Paul McCartney and AI After Now and Then
No—Paul McCartney did not reverse his view of artificial intelligence after “Now and Then.” The Beatles song showed how machine-learning tools can help separate and clarify an existing recording, but McCartney has continued to warn about AI-generated music and the use of artists’ work without permission.
What McCartney said about AI
The apparent change of heart comes from treating two different uses of AI as if they were the same. McCartney supported using technology to recover John Lennon’s voice from a poor-quality demo for a Beatles recording. Later, he publicly objected to proposals that could allow AI companies to use artists’ work without consent or fair compensation.
In a 2023 BBC interview, McCartney described the possibility of AI taking work from songwriters and performers as worrying. He argued that creators should retain control over their work and be paid when it is used. His concern was not simply that computers could make music; it was about who owns and benefits from the material used to train or generate it.
That position is consistent with the “Now and Then” project. The technology was used to make a human performance recorded decades earlier more usable, with the surviving Beatles deciding how to complete and release the song. It was not presented as permission for AI to imitate any musician or exploit their recordings freely.
How AI helped make “Now and Then”
“Now and Then” began as a home demo recorded by Lennon in the late 1970s. After his death, Yoko Ono gave the demo to McCartney, George Harrison and Ringo Starr, along with the recordings that became “Free as a Bird” and “Real Love.” The surviving Beatles worked on the song in the 1990s, but the available tape made it difficult to isolate Lennon’s voice from the piano.
The breakthrough came through audio separation technology developed for Peter Jackson’s documentary project The Beatles: Get Back. Jackson’s team used machine learning to distinguish sound sources within recordings. For “Now and Then,” the process separated Lennon’s vocal from the piano on the demo, allowing the vocal to be cleaned up and incorporated into a new arrangement.
McCartney and Starr then added their own parts, while Harrison’s guitar playing from the earlier sessions was also included. McCartney contributed bass and piano, and the released recording was completed with additional instrumentation and production. The technology helped reveal and isolate a voice already present in the source recording; it did not generate a new Lennon performance.
What the technology did—and did not do
“AI” is a broad label for software that can identify patterns and perform tasks such as separating instruments or voices. In this case, the important point is the distinction between audio restoration and synthetic performance. Separation software can estimate which sounds belong to a vocal or an instrument, then make those elements easier to work with.
That is different from a generative system creating new singing in the style of a person, or producing a voice that sounds like a specific artist without that artist recording it. The “Now and Then” vocal was Lennon’s actual voice from the original demo. The software did not write new lyrics for him, ask a model to imitate him, or fabricate a fresh vocal take.
Audio separation is not magic, either. When voices and instruments overlap in a recording, software may leave artifacts or distort details. The result still needs human listening, editing and musical judgment. For this song, the Beatles’ collaborators used the recovered vocal as part of a carefully finished recording, not as an automatic replacement for musicians or producers.
Why the song does not settle the wider debate
Support for one practical use of AI does not imply support for every use. McCartney could welcome a tool that helped complete a recording made by Lennon while opposing systems that reproduce a performer’s voice or style without permission. These are different questions involving different source material, creative choices and rights.
“Now and Then” also had unusually specific circumstances: an existing Lennon demo, surviving Beatles who could make decisions about the project, and a team working to finish a song connected to the group’s own history. That context is not the same as a stranger uploading a musician’s catalog to generate imitation songs or training a commercial system on recordings without the creator’s agreement.
The broader debate includes copyright, consent, credit and compensation. Laws and licensing arrangements vary, and the ethical questions are not resolved by the fact that a tool can produce convincing audio. McCartney’s comments have focused on protecting creators as technology changes, rather than rejecting all computer-assisted work.
What listeners should take away
The most accurate summary is that McCartney’s position is conditional: he has accepted AI-assisted audio work in a project he helped direct, while warning against uses that undermine artists’ control and livelihoods. “Now and Then” is a concrete example of technology assisting restoration and collaboration, not evidence that he embraced unrestricted AI music.
It is also easy to overstate what happened by saying AI “created” Lennon’s voice on the track. The voice came from Lennon’s recording. Machine-learning tools helped separate it from accompanying sound so it could be used more clearly. That distinction matters when comparing the Beatles’ process with synthetic songs that imitate real performers.
Frequently Asked Questions
Did AI create John Lennon’s voice on “Now and Then”?
No. Lennon’s voice was on his original demo. Audio-separation technology isolated it from the piano so it could be used in the finished recording.
Did Paul McCartney change his mind about AI?
There is no clear evidence of a reversal. He supported a specific use of technology on “Now and Then” and has also criticized AI practices that could exploit artists or take away their control.
Was “Now and Then” made entirely by AI?
No. The song involved the Beatles’ existing recordings and musical contributions, as well as human decisions and production. AI-assisted separation was one part of the process.
Why did the Beatles need AI for the song?
The original demo mixed Lennon’s singing with piano in a way that made the vocal difficult to isolate. Separation technology helped make the voice clearer for the new recording.
Does McCartney oppose all AI-generated music?
His public concerns have centered on the effects of AI on creators, including unauthorized use of their work and loss of income or control. It would be inaccurate to turn those concerns into a blanket statement that he rejects every AI tool.
Related
Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.