Spotif-AI: Data Harvest

As the explosive success of the music streaming business model continually advances, Spotify remains the polished spearhead of the industry. Consistently capturing over 30% of the total music streaming market share year over year, Spotify has accumulated over 600 million users as music streaming surpasses 20 billion in revenue worldwide.

With all that Spotify utilizes to stand out from competitors, they are no less guilty of following significant trends, namely, industry leaders catapulting their business models towards AI incorporation. What could be argued, is that Spotify is doing so to much greater, and more subtle, success than other companies.

DJ X

Upon initial release, I was unsurprised and underwhelmed by Spotify's brand-new AI technology, DJ X. The new AI DJ companion, "voiced" by Spotify's Head of Cultural Partnerships Xavier 'X' Jernigan, appeared to be a poorly contrived marketing stunt amid the AI craze, and served as little more than a decorative casing for Spotify's existing song suggestion algorithms. I have since changed my opinion, as this may be one of the best utilizations of AI for one reason: it is used entirely supplementary to the music.

DJ X is only useable to those who subscribe to a Spotify premium plan, so not everyone has experienced a listening session with X. As an explanation for how DJ X works: he introduces what you've been listening to lately and plays four songs. Then he'll hit you with something more nostalgic and play four songs. Then he'll most likely introduce new music or play something similar in genre to what you would listen to at that time of day/week. If you don't like it, you may skip the song or tap the DJ button to get a more pronounced switch-up.

The simplicity, and X's tranquil voice, create a comfortably dynamic listening experience that doesn't throw you off from whatever mood you are in. The more you listen to and use DJ X, the more acquainted he becomes with your tastes and patterns. The beauty is in the subtleness.

Machine Learning and Data Collection

By now, most people understand that AI requires input to create an output, and it needs a substantial amount of input before it can produce a half-decent output. The strength of Spotify and why it dominates the music streaming service is for two main reasons (as I see it); it has a fantastic UI, and the app's personalization is near limitless. Much of the personalization comes from the suggestions, mixes, and spotlights presented to the listener, all of which come from the AI algorithm's utilization of user data.

While unproblematic in the scope of creating a better listening experience for users, it is apparent that Spotify's data collection vastly reaches beyond the use of improving AI song suggestions.

As reported by CHOICE in May of 2024, the range and quantity of data collected by users of Spotify ranges from app inputs and haptics to data from connected apps and even emotional states. Spotify uses "advanced targeting options" to promote ultra-specific advertisements to users depending on what times of day they are listening, the type of music they listen to at certain times, and the device they are listening on. AI algorithms then use this data to categorize listeners to determine which ads should be shown. Even premium account holders are not safe from this individualized advertising scheme, since 2020 when Spotify introduced ad inserts into podcasts for all account types which can be turned on or off by the podcast creator.

Users can request their data be deleted, but Spotify holds no obligation to fulfill that request. Even if data is deleted, it may still be analyzed beforehand.

Spotify produced a patent in 2021 that served to collect data on age, gender, accent, listening environment, and even the emotional state of users. There has been no implementation of this technology, nor does there appear to be any plans for the streaming giant to employ it in the future.

An image from the patent mentioned above. A simplified blueprint of the technology.

What it Adds Up To

The picture painted by Spotify using AI to harvest vast amounts of data, the utilization of said data, and the technological push for more acute data collection, is one of increased unethical consumerism practices that leave users victims of their habits and humanity.

If every aspect of someone's life can be measured, interpreted, sold, and taken advantage of then there is no end to how much that life can be influenced by those who have the means to do so. A person's uniqueness is turned against them to either sell them something or commoditize them to be sold themselves so more effective advertisements can be pushed on them.

Technology such as Spotify's 2021 patent would permit companies to track almost everything users say or do without pause, and then possibly sell that information to any number of companies (much like Amazon's Alexa, though somehow even more intimate).

As an uncomfortable example, imagine either a horrible or incredible life event in which someone uses music to supplement or medicate themselves to feel better. Suddenly, in that person's unstable or altered emotional state, an algorithm notices the user has reached an emotional extreme and decides to thrust an advertisement at the person. It may feel rewarding for the person to buy something: a little retail therapy or a celebratory purchase. This app has used the power of your emotions for their benefit and has taken advantage of the user who has trusted this music app as a bastion for emotional relief.

The commoditizing of individuals is becoming more and more prevalent. It is rare to find a website, app, or subscription-based business that does not use or sell your information, for better or worse. I will admit that I quite enjoy the use of algorithms to find new music on Spotify, and it is certainly the reason I have not used other music streaming services. Wrapped is always a welcome ending to the year as well of course.

What is left for the consumer to do? To start, you can limit Spotify's ability to track you via your device settings, or by visiting www.spotify.com to access your profile and privacy settings. In your privacy settings, you can turn off Tailored Ads, turn off Facebook tracking, and request all the information that Spotify has collected on you. You can also promote and contribute to platforms such as CHOICE or other privacy protection advocates.

The less that users tolerate AI data harvesting like that of Spotify, the more protection is bestowed upon consumers, and the more companies will be forced to spend resources improving their products instead of improving technology used to record user information.

There is certainly a use case scenario for AI in the music streaming industry, and Spotify has ridden the torrent of AI incorporation well with features like DJ X. It is sadly now the responsibility of the user, and no longer the company, to keep such a tool in check and ensure that it cannot be used maliciously nor without repercussion.