Will You Be Ready?
AI is bringing new people into music creation. Some of them may eventually need musicians.
I have spent quite a bit of time thinking about what AI music could take away from musicians, and I still think that is a very real concern. Lately, though, I have started wondering about another side of this that we are not talking about nearly as much.
AI allows people to get much further into music creation without the traditional skills that used to be required. Someone may write lyrics but not sing. They may hear melodies but not play an instrument. They may understand the kind of song they want but have no idea how to arrange it. Some may have good musical instincts without knowing the first thing about recording.
Others may have no traditional musical background at all.
Until recently, most of those people would have had a difficult time turning the thing in their head into something another person could actually listen to.
Now they can.
That changes something, because once somebody has a song they care about, even if AI helped them get there, they may want more than the generated version. They may want a real singer, guitars, drums or strings they can control. They may want a better arrangement, a proper studio recording or a version they can perform in front of people.
That is where musicians may come back into the picture.
There Is Already a New Kind of Creator
One of the clearest examples is Oliver McCann, who creates music under the name imoliver.
McCann told the Associated Press that he cannot sing, cannot play an instrument and has no musical background. He came from visual design and began experimenting with AI partly as a way to bring his lyrics to life. In 2025, Hallwood Media signed him after one of his AI-created tracks gained millions of streams.
Whatever anyone thinks about whether that makes him an artist, musician, music designer or something else, there is now a person participating in the commercial music business who probably could not have participated in quite the same way before these tools existed.
I don’t think he will be the last.
MIDiA Research is seeing the same change on a much larger scale. Its research says the worldwide number of musicians grew 32.6 percent between 2020 and 2025. Its broader category of music creators, which includes people using generative AI, grew 166.8 percent to 148.7 million. MIDiA describes generative AI as one of the technologies blurring the distinction between consumer and creator.
That distinction between a musician and a music creator may become a lot more important.
I would not automatically call everyone using AI a composer. If somebody writes the lyrics, melody and musical ideas and uses AI to help realize them, composer or songwriter may be completely appropriate.
If AI supplies most of that and the person selects, directs and refines the results, I think we may need different language. Creative director might fit some. Music designer is already being used. There will probably be other names.
Whatever we eventually call these people, there may be a lot more of them.
The Handoff to Musicians Is Already Happening
This was the part that surprised me most. I originally thought we were talking mostly about something that might happen in the future.
We aren’t.
There are already examples of AI being used to take a musical idea far enough that human musicians can take over.
Singer-songwriter Samuel Smith is a particularly interesting case. Smith has Parkinson’s disease, which has severely affected his ability to play guitar. For an instrumental called “Horizon,” he hummed musical ideas into his phone and used Suno and Udio to develop detailed demos.
Those AI recordings were not used as the final recording. They helped Smith communicate what he was hearing to the session musicians who would record the piece. His album involved accomplished human musicians including Jerry Douglas, Alison Brown, Stuart Duncan, Bryan Sutton, Viktor Krauss and Julian Lage.
That is very different from generating a finished song and uploading it to Spotify.
The AI became a way of communicating an idea. The musicians still had to interpret and perform it.
There is also a professional studio example. At a Suno songwriting camp reported by Billboard in March 2026, Grammy-winning producer Om’Mas Keith and other writers generated ideas with Suno and then rebuilt the selected song with musicians. The session included a sought-after drummer, producers and a violinist. By the end of the process, the version heard by the reporter was approximately 90 percent human-recorded.
That gets very close to the production ecosystem I have been wondering about.
Generate an idea. Choose what works. Rearrange it, replace what needs replacing, then perform and record it.
The AI output becomes the starting point instead of the end of the process.
The technology itself is moving in that direction. Suno Studio allows users to work with individual stems and export them as audio or MIDI into a traditional DAW. That makes it much easier for generated material to leave the AI environment and enter a conventional production workflow.
Now there is evidence that a small service market is developing around that handoff.
I found AirGigs job postings from people specifically looking for musicians and producers to rebuild Suno-created demos with human vocals and instruments. One posting described a songwriter who had written the lyrics and melody, used Suno to create a rough arrangement, and wanted a full-band human recording. The person said there could be as many as six additional songs if the first collaboration worked out.
Another creator was looking for an ongoing production relationship to turn AI-assisted demos into human-produced recordings.
Musicians are responding from the other direction too. Producers, singers and session musicians on AirGigs, SoundBetter and Fiverr are already advertising services for rebuilding Suno and Udio songs with human vocals, real instruments, new arrangements and professional production.
I don’t want to make that sound bigger than it is. A few marketplace listings are not proof of an enormous new industry.
They do prove that the transaction already exists.
Someone has an AI-assisted song. Someone else has the musicianship needed to take it further. Money changes hands between them.
That is worth paying attention to.
A Generated Song Is Not Automatically a Live Act
This is where performing musicians may have an advantage that is very different from what happens on streaming.
An AI system can create an audio file. That does not automatically create a band or someone who can stand onstage and sing the song. It does not create a drummer who can follow a singer stretching a phrase, musicians who can recover when something goes wrong, or the chemistry that develops between people who rehearse and perform together.
A generated recording also does not change from night to night.
We have already seen an interesting demonstration of that gap. Composer and producer Adrian Younge took “Through My Soul,” a successful AI-generated soul track, arranged it for his Midnight Hour band and singer Loren Oden, recorded a human version and performed it live in Los Angeles. He later decided to keep it in his touring set.
I find that fascinating.
Whatever you think of the original AI song, the generated recording became material that a group of musicians could interpret. Musicians have been doing that with compositions forever.
The unusual part is where this particular composition came from.
People are also still spending heavily on live music. Omdia reports that worldwide concert and festival ticket revenue exceeded $40 billion in 2025 and projects it will exceed $50 billion by 2030.
That does not mean people will suddenly pay to see every AI creator perform. Most of them certainly won’t. It does tell me that recorded audio and live musical experience remain very different markets.
AI may be able to flood one much more easily than the other.
This Is Not a Promise That AI Will Save Musicians
I want to be careful here. I am not saying AI is secretly going to save everyone’s career.
I don’t know that.
AI is also capable of taking work away from musicians, singers, composers and producers. CISAC commissioned an economic study that estimated 24 percent of music creators’ revenues could be at risk by 2028 under current conditions, particularly as AI-generated material substitutes for human work in areas such as streaming and music libraries.
The amount of generated music is becoming enormous. Deezer reported in July 2026 that it was receiving around 90,000 fully AI-generated tracks per day, with AI music exceeding half of its new daily uploads at the June peak.
Most of those songs are not going to hire a guitarist. A person generating hundreds of tracks to fill streaming accounts probably has no interest in spending money on musicians.
That is not the market I am talking about.
I am interested in the person who creates one song that suddenly matters to them.
Maybe they wrote the lyrics about their wife. Maybe it is something they have been carrying around for twenty years. They may hear a melody but have never been able to play it. AI may help them discover that they are actually good at writing songs even if they are terrible at performing them.
Or maybe they simply create something they like enough that the cheap and easy version is no longer good enough.
That person is different.
That person may want help.
The Opportunity May Be in Finishing What AI Starts
If this market grows, the customer may not come to a musician with a chord sheet and a clean demo.
They may hand over a Suno link and ask:
“Can we make this real?”
Answering that question requires more than knowing how to play an instrument.
A musician may have to figure out what the AI actually did, transcribe a part or simplify something that sounds fine in generated audio but is awkward for a human performer. A singer may need to replace the vocal. An arranger may need to work out which strange little accident in the AI version is actually one of the things the creator likes most.
Someone may have to turn that generated recording into an arrangement five people can perform onstage.
That moves musicianship into another category of value.
Not simply, “I can play guitar.”
More like, “I can understand what you’re trying to make and help you get there.”
That is much harder to automate.
Will You Be Ready?
If I were a working musician interested in this kind of opportunity, I would at least start preparing for it now.
- Learn how to work from AI demos, stems and reference tracks without expecting the client to understand normal studio language.
- Consider offering specific services such as AI-demo reconstruction, human vocal replacement, live-instrument replacement, arrangement development or converting a generated song into a live-band arrangement.
- Build one or two examples showing how you can take a rough or generated reference and turn it into a human performance.
- Make remote collaboration easy. Clear file delivery, stems, charts, revisions and communication may matter almost as much as the playing.
- Be clear about rights, credits and AI boundaries. Ask what material the client created, what tools were used and what they actually own before agreeing to release or perform something.
- Decide your ethical line before somebody pays you to cross it. Rebuilding someone’s original AI-assisted demo is very different from being asked to imitate a living singer or copy another artist’s identity without permission.
The rights issue should not be ignored. The U.S. Copyright Office says AI-assisted work can still receive copyright protection where sufficient human-authored expression exists, including human creative arrangements and modifications. Prompts by themselves generally do not provide sufficient human authorship.
Hiring musicians to rerecord something does not automatically resolve every ownership question surrounding the underlying generated material.
Musicians entering this kind of work may eventually need to become smarter about rights and provenance, not just better at recording.
Maybe the Job Changes
AI may create a much larger population of people who have musical ideas but do not have all the skills needed to finish them.
That used to be a very high wall. People needed access to musicians just to discover what their ideas might sound like.
AI lowers that wall. Someone can experiment privately, cheaply and repeatedly until they have enough of an idea to show another person.
I understand why musicians see that as threatening. Some of the work that once happened between musicians may now happen inside the machine.
But I can also see the other possibility.
People who never would have approached a musician before may eventually arrive with something in their hands.
Not a finished record.
A blueprint.
The part about partnerships is where this really gets interesting to me. We may be looking at a new kind of production ecosystem. Someone comes up with the song or concept using AI, but when it is time to perform it, record it or take it onstage, they may still have to bring in musicians, singers, arrangers and producers.
Musicianship does not disappear. It becomes a specialized skill people may have to seek out and pay for.
Some creators will be perfectly satisfied leaving everything inside the machine. Others won’t. They will want singers, producers and arrangers. They will want guitarists, drummers, bass players, keyboard players, string players and people who know what to do when generated audio has to become something human beings can actually perform.
I don’t know how large this becomes. I don’t know whether the work created will outweigh the work AI takes away.
Nobody knows that yet.
But the category already exists. People are already doing this work.
Maybe the musicians who prepare for that and position themselves well will be the ones who stay busy.
The question is worth asking now rather than later:
Will you be ready?