māchine · music-streaming

Wired's reviewer finds AI DJs on YouTube Music and Spotify get facts and names wrong

A Wired reporter used both services' synthetic hosts for months and quit them. The voices sound human enough. The commentary is the problem: filler, errors and trivia that does not match the music.

Both of the big streaming services now place a synthetic voice between songs. YouTube Music tests AI music hosts through its YouTube Labs program. Spotify has had an AI DJ since 2023 and expanded it on November 10, 2026. In a first-person review for Wired, a software reporter describes using both and then giving them up.

The reporter does not object to how the voices sound. Their complaint is about what the voices say. The commentary repeats empty phrases, and some of it is wrong. That matters because both products now do work that used to belong to radio hosts and playlist editors: picking the next song and telling the listener why it deserves attention.

What the two products do

YouTube introduced YouTube Labs in September 2025 as an initiative for exploring AI on YouTube. AI music hosts were its first experiment. YouTube describes them as "AI music hosts designed to deepen your listening experience," sharing stories, fan trivia and commentary about the music a person listens to. Access is limited to a set number of US-based participants. Wired refers to the feature as Beyond the Beat and says the host breaks in at random for about 30 seconds between tracks, setting up the change in mood and giving background on the next song.

Spotify's DJ goes further. According to Wired, it also picks the songs, using a listener's history and their requests. Spotify's November announcement describes DJ as a personalized lineup of music with context and commentary tailored to each listener's taste. Pressing the DJ button switches the lineup to a different genre, artist, mood or set of recommendations.

The update added nine voices alongside the original voice, DJ X. Seven are in English: Casey, Tess and Lane (US), Sam (Australian), Nell (UK), Declan (Irish) and Nisha (Indian). Two are in Spanish: Nico and Fabi, both Mexican. Listeners can now turn off the commentary between songs. Brief acknowledgments still play when DJ starts or when the DJ button is pressed. The new voices and controls are for Premium subscribers in markets where DJ is already offered. Spotify says 80 percent of listeners come back to DJ week after week.

Where the commentary failed

Wired gives two examples from YouTube Music. In one, during a workout playlist, the host followed a Slayyyter track with roughly 30 seconds about Charli XCX's song "Rock Music". It said the song deals with nerve damage the singer sustained while touring in 2024 and 2025, and that she reclaims "the exhaustion of a weary party girl as a badge of honor." The host also mispronounced the artist's name. Wired's transcript renders it as "Charli XXC." The reporter calls the passage "a lengthy paragraph of platitudes with little substance."

In the second example, the host talked at length about the meaning of an Olivia Rodrigo remix featuring Chappell Roan. It called the track a "structural interpretation" exploring the "nuances of queer dating." Then YouTube Music played the original version, which does not include Roan. The commentary was written for a different recording from the one the listener heard.

To compare, the reporter upgraded to Spotify Premium and listened to the introduction clips for the eight English-language voices. Several used filler words such as "you know" and "like" when describing upcoming songs. The reporter says that knowing a voice had been built to use filler in order to sound more human produced an uncanny-valley reaction. Overall, they found Spotify's commentary tighter and easier to tolerate than YouTube's. Neither was good enough to keep using.

Voice is solved, content is not

The review separates two problems that product announcements tend to treat as one. On the reporter's account, synthetic speech has reached the point where both companies' voices are "decent and sound convincingly human." The commentary has not caught up. The YouTube host produced fluent, confident text that was attached to the wrong recording and got a well-known artist's name wrong. The Spotify voices were made to sound casual by copying human hesitation.

The two companies are responding in different ways. Spotify's November update gives listeners a way to make the DJ talk less, and it offers a wider range of voices. YouTube's hosts are still a limited experiment. Wired notes that both features are optional. For listeners who want to avoid them, the simplest response is to leave them off.

What it means for work

These products automate two jobs that used to be done separately: choosing the running order, which programmers and playlist editors did, and introducing each track, which radio hosts did. Spotify's DJ does both. YouTube's hosts so far only add commentary to music the listener has already chosen. The Wired examples show which part breaks first. Generated trivia is cheap to produce, but nobody checks it against the track that actually plays, and a human host would not mispronounce Charli XCX or describe a featured artist on a version that leaves her out. A human DJ's value came partly from being accountable for the facts and partly from knowing when to stay quiet. Spotify's new option to switch off commentary suggests the company has noticed the second part. It is still unclear whether fact-checking, the first part, will be built into these systems or left to listeners to catch.

Sources

This article was researched, written and illustrated by AI, and published without human editing. Spotted an error? Let us know.