stub New Study Sheds Light on 'Algorithmic Fatigue' - Unite.AI
Connect with us

Artificial Intelligence

New Study Sheds Light on ‘Algorithmic Fatigue’



A new study conducted by strategic consultancy Alice Labs and global technology firm Reaktor sheds light on what researchers call “algorithmic fatigue.” Algorithmic fatigue takes place when individuals spend long periods of time doing things like browsing streaming services. In these moments, AI systems can often fail in their duty to satisfy the expectations of users if they don’t deliver. This can often leave consumers feeling fatigued. 

The new study identified three different types of AI interactions: 

  • Passive: Users want to remain passive towards an algorithmic system.
  • Guiding: Users want to guide an algorithmic system.
  • Collaborative: Users want to collaborate with an algorithmic system.

The type of AI interaction a user selects depends on the situation and what they want or need, and it is also affected by previous experiences with smart technology, such as algorithmic fatigue when a previous system failed to deliver those needs. 

According to the researchers, AI is becoming more than just technology, and brands are beginning to realize the importance of how users experience and interact with algorithms. 

Kirsi Hantula is one of the researchers at Alice Labs. 

“While useful in many instances, algorithms continue to be limited by their machine-ness: they cannot predict when users are having a bad day and need something lighter to watch, nor are they capable of understanding the subtle and varied ways in which users’ tastes evolve and expand over time,” Hantula says. 

Because various AI-powered devices have been around for over ten years, many users are already aware of how to tactically navigate imperfect systems. This means that users themselves have already begun to fight against algorithmic fatigue. For example, one might limit an AI-powered device to its basic functions, such as using a voice-activated assistant for simple tasks. 

Recommendation Algorithms

The researchers also found that more consumers are now opting out of recommendation algorithms, moving more towards external influencers to curate content. External influencers, or lighthouses, are other individuals who share similar interests, and these individuals are often more reliable than the algorithms. 

The researchers say that companies should find ways to combat algorithmic fatigue if they want to establish stronger relationships with their consumers. 

Olof Hoverfält is Principal Consultant, Strategy and Business Design, at Reaktor. 

“It’s not so much about reinventing AI, the AI is working fine,” says Hoverfält. “It’s about creating another layer on top of that system, something that allows for quicker, more refined human interaction between the user and the algorithm. It’s about parity: granting the user equal agency over decision making, allowing them to choose and change when they want to be actively involved in the process or just passively guided through it.”

“For us, this is also a matter of ethics. We think that AI systems that intimately intertwine with people’s lives should be designed in a way that celebrates human versatility and establishes users as positive and creative agents in algorithmic decision-making,” says Hantula.

The new study is part of a two-year collaboration between Alice Labs and the University of Helsinki’s Center for Consumer Society Research. It was funded by the Foundation for Economic Education.

Alex McFarland is a tech writer who covers the latest developments in artificial intelligence. He has worked with AI startups and publications across the globe.