Behavior Model - Spotify as an Example

Spotify has 626 million monthly active users, with 246 million being paying subscribers. How did Spotify become the global leader in music streaming?

Through complete customer focus. Spotify knows which music Laura wants to listen to. It observes her behavior, understanding which tracks she might enjoy. Spotify maintains knowledge about Laura's favorite music, her recommendations to friends, the musical tastes of her friends, the characteristics of available music like genre, artist, tempo, intensity, key, and even the environment and timing of music playback. In other words, Spotify understands Laura's music needs, probably better than Laura herself, and knows the global music offering, certainly better than Laura does.

Laura’s World (consumer´s view)

For instance, Spotify plays “Another Brick in The Wall” for Laura. Laura perceives specific notes, rhythm, tempo and volume. These attributes fulfil needs that can be categorized under the need for health. Her brain releases dopamine and produces a feeling of well-being, even goosebumps. The action of playing the song, the perceptions, needs and pleasant feelings form her knowledge of the song, which is all around positive, prompting her to play it again.

Daniel Ek’s World (supplier´s view)

From the perspective of Daniel Ek, Spotify’s CEO, it might mean: choose the action "play a preferred song," observe the user’s duration of retention and recommendations, evaluate the fulfillment of Spotify’s needs—listening time, repeats, referrals, loyalty, ad revenue, and subscription fees. The impact gives Daniel Ek satisfaction since it helps him meet financial goals and strengthen his standing with shareholders.

Short- vs. Long-Term Need Fulfillment

If services satisfy people's needs, they improve quality of life, and everything develops for the betterment of humanity—in theory. However, what’s critical for life quality is whether services meet customer needs in the short or long term.

It is concerning that service providers today almost exclusively aim for hedonia—the quick gratification of needs. However, listening to music can hinder focus during study or work, can replace conversations, within families for example, and prolonged exposure to loud music can damage hearing. What is good for people in the short term may not always be beneficial in the long run. Should service providers prioritize hedonia or eudaimonia, the long-term well-being of consumers?

Stakeholders

Digital services like Spotify, YouTube, Microsoft Office, or Amazon have been building behavioral models of users for years through knowledge graphs, continuously refined by deep learning. They tailor their offerings to customer needs. Academia supports this through research on recommending engines.

Whenever  you, like Laura, consume music on Spotify, you are one stakeholder. But other stakeholders include artists, shareholders, music labels, etc. The needs of the user, you, and the interests of other stakeholders do not always align. The consumer’s desire for a satisfying life and shareholders’ desire for profit often conflict. This tension between human needs and economic interests is reflected in academia.

Business Engineering vs. Life Engineering

Business Engineering and Life Engineering are two distinct scientific approaches, often pursuing divergent goals. Business Engineering focuses on shareholder interests, driving actions that directly lead to sales or indirectly influence consumers. Put simply, it caters to the need for capital, prioritizing immediate and short-term satisfaction of needs over users’ long-term well-being. Hedonia determines consumer behavior and is thus more critical to Business Engineering than eudaimonia. Spotify fulfills the need for musical stimulation for instant comfort, without concern for disruptions to work or potential hearing damage.

However, if machine intelligence is meant to benefit humanity, the focus must be on long-term well-being. Business Engineering starts from a business perspective, while Life Engineering starts from human life. Life Engineering seeks a behavior model that delivers high quality of life - eudaimonia - over time.

Life Engineering’s Behavior Modeling

The rudimentary needs model of Life Engineering takes into account 13 human needs to model behavior accordingly. Massive data collections, AI algorithms, and powerful AI processors enable increasingly effective models. If we understand the diverse needs of humans in relation to their behavior, we can use hints, recommendations, and coaching to improve individual life quality.

Quality of Life

Machine learning, for instance, could deduce emotions from biometric data like heart rate variability from Apple Health, prosody, i.e.  tone of voice, from other sensors, and swipe patterns on an iPad. This enables machine intelligence to assess how actions impact life quality and serve as a life coach for consumers.

Digital Life Coach: Utopia or Imminent Reality?

Will AI eventually understand your needs better than you do? Can it derive behavioral rules from everyone’s needs? If the answer to these questions is yes, then AI could establish new ethical values for the benefit of individuals and society. A digital life coach could guide people.

This is technically and socially a huge challenge. It requires, first, the willingness of service providers and users to let the life coach use sensor and usage data. Second, a machine learning concept must be developed that integrates and standardizes multimodal data sources and refines the rough needs model of Life Engineering. Once these hurdles are overcome, third, Apple and other service providers would need to find a commercial interest in investing in this AI. They must believe in the potential monetization of such capabilities. Fourth, an international regulatory framework is needed to measure the contribution of service providers to the life quality of all people. Only then can economic incentives and societal interests align, motivating services like Spotify to develop and offer life coaches.

Laura learns every day, around the clock, through her sensors—eyes, ears, nose, mouth, muscles, skin—and through media with text, images, audio, and video, which actions make her happy or unhappy. Why shouldn’t a machine also be able to do this if it can measure feelings, such as happiness and sadness, and build a behavior model?

This may sound utopian, but elements of it are already becoming reality. Digital coaches are emerging in the workplace, like Microsoft Viva, in health, like Apple Health, or in financial planning, like in eBanking. It’s easy to imagine how these coaches might one day merge into a comprehensive life coach. Life Engineering’s task is to guide this towards improving life quality for everyone.

We Have a Choice and Must Decide.

Should the behavior models of service providers narrowly pursue the need for capital or should they serve all human needs? This will determine whether machine intelligence leads to the benefit or detriment of people. In the end, it is not primarily the regulators who determine the direction, but the most technologically advanced. If we don't change anything, the battle for the most powerful AI between the technopolists and countries will continue, whatever the cost. In a few years, the result could be an AI whose only goal is to develop its own intelligence and power, regardless of what happens to humans.

Questions

• Why do you use Spotify or a similar music service?

• What health apps do you use?

• Who will own the most powerful behavior model in five years?

Behavior Model WIX en 01 hoe.pdf

Download PDF • 224KB