Aside from using AI to generate code , essays, pictures , and podcasts, you can also now produce music and music videos. In fact, AI music tools are so accessible and capable that you've probably already heard their creations on Spotify, TikTok, and YouTube without realizing it. This is all deeply impressive from a technical perspective, but I'm convinced AI music is a huge mistake: It sucks all the creativity and joy out of the process.
How Do You Make Music (and Music Videos) With AI?
Simply search for an AI music generator, and you will find countless such services. I used Suno , but its features and results aren’t especially unique. Making a song largely involves clicking the Create button. You can give the AI as much or as little information as you want. For example, Suno let me choose a genre, generate lyrics (though I chose to write them myself), pick a vocal gender, and even upload an audio clip for reference, all at no cost. Then, seconds later, it produces a song. If this all sounds jarringly easy, that’s because it is, unfortunately.
Making music with AI doesn't require prompting, and the only actual creative work you need to do is pick a genre or the type of sound you want. I opted for a country pop song, and it sounds very much like something you might hear on the radio or on a Spotify playlist. You might raise an eyebrow at certain pronunciations, which I could fix by spelling them out phonetically in my lyrics, but otherwise, it's hard not to find the technology impressive.
But why stop there? You can also make music videos with AI in a variety of ways. For example, you can prompt Gemini’s Veo 3.1 AI video generator to create a video and then add your music of choice as a backing track. However, dedicated AI music video generator services also exist. I tried neural frames , which gives you an incredible amount of control over the music video generation process, down to the level of frame-by-frame adjustments.
I provided neural frames with a song I generated with Suno, and then I let the AI choose a concept and style for my video. It then presented me with characters, a story description, a slider to choose how much I want my video to focus on vibe over story, and a storyboard, complete with images and descriptions for each scene. Finally, I chose a model from a list of options, and neural frames started generating dozens of clips it stitched into scenes to make my video.
Behold: My Pop-Country AI Anthem and Lo-Fi Masterpiece
The soon-to-be classic (above) is just the beginning of what you can do with neural frames, considering you can also generate lyric videos or even ones in which AI characters lip sync songs. While the depth of neural frames is impressive, results are a lot more variable, considering the available models aren’t at the cutting edge of AI video generation. However, with enough tweaking and spending ( credits cost money ), good results are possible.
I also generated a chill, lo-fi song with Suno (below). This required next to no work, since all I did was select my genre.
AI Music Is Everywhere and Sometimes Impossible to Spot
Outside of picking out mispronunciations (you can hear one in my song above when the AI tries to sing the word ‘AI’ at certain moments and doesn’t quite get it right) or listening keenly, the reality is that you likely can't distinguish most AI music from human-created music. Although some streaming services label AI-generated music , not all do, and they continue to face a torrent of AI music uploads .
The second video above should sound familiar to you if you listen to lo-fi music, as I do sometimes while working, courtesy of Lofi Girl . However, Lofi Girl notes that humans create all of the music and visuals in its videos. However, many channels don’t have those scruples. When I search for lo-fi beats on YouTube, Lofi Girl is the first result, but a similar channel, called The Japanese Town , is the second.
The latter channel doesn’t mention AI anywhere, simply noting that The Japanese Town team creates all the artwork and music in its videos. However, it's all almost certainly the work of AI. The artwork in its videos includes distorted or nonsensical elements, and its streams have almost identical runtimes. Meanwhile, its thumbnails are nearly the same when it comes to style and composition, yet they remain distinct, a hallmark of AI image generation . Unless The Japanese Town is one of the most prolific collectives ever, creating hundreds of hours of unique music without so much as a footprint on the wider web, it’s all AI-generated.
I could easily wind up listening to AI-generated music without noticing it, which is why I steer clear of lo-fi music playlists from sources I don’t trust. But if I can’t always tell the difference, what does it matter where the music comes from?
Listening to AI Slop Is Wrong On Many Counts
It’s one thing if you’re a seasoned music producer who wants to experiment with sampling some AI-generated vocals in a new song, even if that comes with its own set of ethical considerations. It’s another thing if you want to make a meme song to send to your friends. But knowingly spending your time listening to music that's wholly AI-generated is just gross.
That kind of AI music is thoughtless slop that requires zero creative expression and next to no effort. It’s a shallow imitation of one of the few things that makes life actually worth living (art), and it exists to fill the pockets of faceless ghouls who prompt an LLM instead of actual artists.
You must also consider the potential legal ramifications of profiting from AI-generated music. PCMag's Jamie Lendino , executive editor of reviews and longtime audio producer, points out that "LLMs have scraped enough music and violated enough copyrights to present a reasonable facsimile of music in different genres" but notes that you "can never be sure it didn’t copy a melody, which means you really shouldn’t produce anything and present it as your own, lest you get sued."
Beyond the legal issues, PCMag's Angela Moscaritolo , managing editor of consumer electronics, finds it concerning that people are using AI-generated music to spread misinformation and conspiracy theories. "Following the Artemis II mission, a friend sent me an Instagram Reel suggesting NASA faked the Apollo program Moon landings, featuring a song called 'Radiation_Fire' by an artist dubbed the Deep State Ramblers," she says. "Based on the singer's voice, I initially thought it was a Dolly Parton song, but a quick Google search proved it was AI." That Instagram post had over 53,000 likes as of publication.
Even if none of the above gives you pause, AI music is usually just bad. Sure, not much might separate an AI-generated chill, lo-fi beat from a human-created one in terms of sheer musical quality, but listen to the country-pop song I made with AI and tell me it doesn’t suck. The reason why you might mistake that song for something authentic is that it's generic by design. The underlying LLM trains on countless country-pop songs, which is how it creates a song that sounds exactly like others in the genre.
Don’t Go Too Far With AI Music Generation
If you’re in a garage band just getting off the ground and you want to upload your music to YouTube with more than a still image in the background, an app like neural frames is not only technically impressive but could also be useful. (Of course, aspiring videographers might rightly disagree.) AI music generators are just as impressive from a technical perspective, but I can’t in good conscience recommend using them for anything beyond a silly experiment.
PCMag and Yahoo may earn commission from links in this article.
