(January 12, 2026) Two months ago, we posted an article that created quite a stir, as we told the story of a convincing song submission to SoulTracks by a fresh young artist who turned out to be completely A.I.-generated. It was an existential moment for us, as the ongoing discussion of A.I. in music became very real here (it led us to create a policy on covering A.I.-generated music). Since then, I would estimate that 10-20% of the new music submitted to us for review is A.I. based.
So, I spent a good portion of my Christmas holiday digging into the world of A.I. music firsthand, using a popular A.I. music generation program to both modify my own songs and create new music from whole cloth. And the results surprised me.
I began by uploading some songs my brother and I wrote and recorded years ago – sonically poor recordings of our amateur music. The goal was to freshen them up and provide different arrangements, purely for fun. The results were fast and shocking. A muffled soft rock song of love gone bad now became a crisp, horn filled, jazzy number, and an murky midtempo pop song now leaped from the speakers as a dance cut, filled with handclaps and orchestral breakdowns. It was exhilarating to hear the new aural clothes on my long lost songs. But when I played them for loved ones and friends, the response was pretty uniform: They liked the imperfect, heartfelt originals better than the slick, soulless updates.
That reaction presaged my own conclusions as I leaned into full A.I. music generation tools. I created a suite of instrumental jazz fusion numbers, as well as “hard listening” songs in a variety of genres. All I needed to do was provide prompts about lyrical theme, musical style and some instrumentation choices, and voila, a new song was created on the spot.
What the A.I. tool spat out was uniformly catchy and immediately appealing. But each song also gave me this feeling of musical déjà vu – I was certain I had heard it before, even though it had just been created. And it turns out, that is the siren song of A.I.-generated music. Each time, it delivers to you a song that sounds oddly familiar, and if there is one thing I’ve learned about music discovery after all these years, it is that there is short-term comfort in the familiar.
An AI music program is trained on vast collections of existing compositions, including many of the most popular songs of our time. As a result, it can generate new music that convincingly evokes the style and feel of those works without directly copying them. So, a song that mimics something I’ve loved in the past, but is not strictly a remake, is actually more immediately attractive than something that pushes boundaries — that A.I. generated song that vaguely reminded me of Roy Ayers-style jazz/funk or lush 70s Philly soul easily won me over on first listen.
But here is where the surprise came: Those new songs that were pure dopamine for me on first listen had no staying power, because there was nothing new, interesting or real about them. While they seemingly hit all the right notes, ultimately the lack of depth made them downright boring after a few listens. And it wasn’t just one song that did it; every single song I created had a half-life of about 4 listens. Music is meant to evoke feelings, and over time the ersatz R&B or jazz numbers simply didn’t move me, in the same way it wouldn’t move me to hear a programmed Siri say “I love you,” but it means everything if the same words are said with feeling by my wife.
I will admit that I have worried about A.I. taking over the creation of future music, but this experiment made me more hopeful. There will certainly be types of music (particularly ambient or background music) where A.I. will continue to grow – and because of its near free cost, perhaps dominate. And, just as synthesizers or autotune did in the past, it will undoubtedly be used as a new tool to add instrumental elements or sweeteners to otherwise human-created music. But almost by definition, A.I.-generated music is stuck in the past, imitating the prior work of others rather than moving music forward. Musical innovation is still reserved for humans, and innovation and the expression of real, relatable emotion will continue to be what excites us and provides the meaning and connection that drives music to its next place. So, the arrival of A.I. means music will continue to change, as it has consistently since cave dwellers were pounding sticks on hollow logs, but we may find that the next stage is less seismic than we fear.
By Chris Rizik
