J Mascis says AI in music is obvious but not the future

Mascis said the clip came from a young fan. In his words, “some kid did it,” and he thought it looked cool. He added that he doesn’t really do much with videos, which suggests he didn’t think hard about how it was made.

He trailed off several times while answering, then admitted the obvious problem: it was, he said, “a slippery slope.” That is a change from simply letting the video stand. It doesn’t sound like regret so much as a late recognition that using AI imagery, even casually, puts an artist on one side of a heated debate.

The video drew criticism at the time, given how tied he is to analog sound. His lukewarm explanation is unlikely to settle that, but it is the first time he has addressed it publicly.

Mascis doesn’t believe AI has a place in making music. He also isn’t pretending it will go away. His view is that it’s here whether people like it or not, and that everyone is still learning how to deal with it. He added a dry hope that it won’t kill us all.

That is the gap in the headline: AI being everywhere is obvious to him, but he doesn’t see it as how songs should be made. It’s a practical position rather than a campaign. He isn’t calling for bans or boycotts. He simply won’t use it in the studio himself.

His studio habits explain the attitude. Mascis calls himself a “tube purist,” meaning he loves the warm sound of vacuum-tube amps and old gear. But he admits that purity is often inconvenient, so he uses whatever works. On the new Dinosaur Jr. record, There Near, that meant recording in Pro Tools, the standard computer recording software, while still mixing and mastering to tape.

The band also kept things loose. There were no eight-hour studio marathons and no endless file swapping between members working remotely. Mascis brought in finished demos, and the group started recording right away, even while bassist Lou Barlow and drummer Patrick Murphy were still learning their parts.

His approach to gear design is similar. For his signature Fender guitars, he cares most about the neck and big jumbo frets, and wants the rest kept simple and fairly cheap so players can swap parts as they like. Asked about the band’s staying power, he joked that they have “nothing better to do.”

In other words, his line isn’t “no computers.” It’s that technology should serve the playing, not replace it.

The AI side is trying to look friendlier. Suno has promised a way to detect songs made with its tools, and it is rolling out download caps meant to slow the stream of AI tracks onto streaming services. It is also pushing Suno Studio, a version aimed at musicians that gives them much finer control over what the AI produces.

Not every court is buying the argument that training is harmless. A German court sided with GEMA, the country’s music rights society, ruling that Suno broke copyright law by training on its members’ songs and ordering it to pay damages. For artists wary of AI, like Mascis, that is a sign the rules are still being written.

For fans, the clearest takeaway comes from how Mascis talks about live shows. He thinks audiences are hungry for something real. Dinosaur Jr. plays without in-ear monitors or click tracks, and Mascis still stands in front of three full amp stacks, even in a club holding 280 people. He complained that sound engineers now create a band’s sound rather than capture it.

That hunger for the real thing is colliding with a fast-growing AI music business. The leading AI song generator, Suno, has been sued by major labels since 2024 over training its system on their recordings without permission. Warner Music Group settled and signed a licensing deal. BMG, another big music company, struck a deal that requires its artists and songwriters to opt in before their work is used, and pays them when they do.

Universal Music Group and Sony Music are still fighting Suno in court. Evidence gathering in that case is set to run through mid-2027, after which a judge will decide whether the training counted as fair use. The labels want data from distributor DistroKid on how many AI-generated tracks reach streaming services, to show machine-made songs compete with human ones. Apple Music has handed over details on how it detects and labels AI music, and is resisting a deeper deposition.

For musicians, that means the choice Mascis made casually, letting AI near your work, is becoming a business and legal decision. Some stars, such as Madonna, have already refused to let their music be used for AI training.