The Algorithmic Experience (AX)

Author(s): Alexandra Lebleu

Originally published on Towards AI.

Users are aware when they tap a button or scroll through an app, but few realize that behind the scenes, algorithms are quietly shaping their experience.

A survey from Norway (a country with near-universal internet and smartphone adoption) found that 41% of people were completely unaware of algorithms’ role in their digital lives. From recruitment platforms shaping career opportunities to dating apps influencing relationships, their influence is greater than we often realize.

Human-Computer Interaction (HCI) and User Experience (UX) research have spent decades perfecting interfaces, refining accessibility, usability, and navigation to near-perfection. Yet traditional HCI and UX frameworks rarely consider how users perceive, interpret or are influenced by these technologies that increasingly shape our digital experiences.

In response, a growing body of research has started to explore algorithmic experience or “AX”, focusing on understanding user–algorithm interactions.

The spectrum

Consider the features we interact with daily: finding a contact in our phone or hitting Ctrl+Z in a text editor. These systems rely heavily on algorithms, yet we experience them as predictable, reliable tools. Users face no friction.

These are deterministic algorithms. Given the same input, they always produce the same output. Search for “Mom” in your contacts, and Mom appears. Hit undo, and the last action reverses. Outcomes are consistent, so users rarely question why something happened or feel uncertain about the process.

In contrast, many of today’s digital products are shaped by opaque complex algorithms. Their outcomes can be hard for users to predict or fully understand, creating potential friction and experiences that may feel unpredictable from the user’s perspective. They are black-box algorithms.

This unpredictability introduces a new dynamic between users and technology, one that traditional HCI and UX frameworks doesn’t capture.

The algorithm hates me

If you’ve scrolled social media lately you’ve probably seen those “the algorithm hates me” posts. Behind the memes lies a reality, we inherently assign emotions and motivations to lines of code.

This isn’t new. Since the 1990s the Computers Are Social Actors (CASA) paradigm has shown that we instinctively treat computers like social beings, even forming emotional attachments. We unconsciously apply social rules, expectations, and behaviors to computers and other technologies.

We anthropomorphize technology. Which can make unintended outcomes feel personal, even adversarial.

Our relationship with algorithms is complicated. We rely on them daily, often without realizing it, craving their convenience while simultaneously distrusting their results.

Users often exhibit algorithm aversion when they know algorithms are involved, even when algorithms clearly outperform human judgment. For example in 2023 about two-thirds of Americans said they would not want to apply for a job if AI were used to help make hiring decisions.

Unpredictability, trust, and misaligned goals

Traditional UX is built around predictability: tap a button, get the expected result, aligning with user expectations and satisfaction. Products driven by “black-box” algorithms can disrupt this logic and users’ mental models, as they often pursue goals such as engagement or revenue rather than users’ satisfaction.

For non-tech savvy users especially, this variability can come across as unreliability, weakening trust in technology. Identical inputs may yield different outputs, leaving people uncertain about which results to rely on, or whether they should rely on them at all.

Influence without awareness
There is also an ethical dimension to consider: algorithms can shape not only what we see but also how we think. Black-box models can be susceptible to bias, as they may reproduce human biases present in their training data or embedded in their design.
These systems can subtly influence our preferences through repeated exposure to curated content, often without our conscious awareness. They create echo chambers that reinforce existing beliefs while filtering out diverse perspectives, thereby strengthening biases and limiting critical thinking and intellectual growth.

Consider this: fewer than 1% of people venture beyond Google’s first page, giving enormous control over the information we encounter.

Path forward

Early AX research aims to design more human-centered experiences. By emphasizing transparency, user control, and literacy, we can build digital experiences that are clear, equitable, and genuinely empowering for all.

Digital Literacy
Studies show significant gaps in algorithmic understanding, particularly among older adults, low-income individuals, and those with less education. These groups may trust outputs blindly, even when results are biased. Universal design principles and targeted education can help ensure everyone benefits from technology while preventing unfair exclusion.

Transparency & explainability
Help users understand what data is used, why decisions are made, and the reasoning behind them. This doesn’t require exposing complex code, just being more transparent about the process. Transparency allows users to spot potential errors or biases in algorithmic decisions.

Research shows that explainability fosters trust: Stanford students who saw how their grades were calculated trusted the algorithm more, even with lower scores.

The field of Explainable Artificial Intelligence (XAI) explores methods that allows human users to comprehend and trust the results and output created by machine learning algorithms.

User control
Provide meaningful ways for users to influence or override algorithmic outputs. Even minimal interventions such as TikTok’s “Not Interested” button can significantly restore user agency and system trust.
More sophisticated approaches include letting users adjust algorithmic parameters directly.
If a black box model makes wrong decisions or produces inaccurate or harmful outputs, it can be difficult to adjust the model or provide feedback.

As we become more dependent on digital products and algorithms become increasingly ubiquitous, the field of Algorithmic Experience (AX) will help us improve these interactions and examine how these systems shape choices, trust, and behavior.

Collaboratively written by Yerko Ortiz Mora and Alexandra Lebleu

• Alvarado, O., & Waern, A. (2018). Towards algorithmic experience: Initial efforts for designing algorithmic systems. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems.
• Dang-Pham, D. (2020). An overview of Algorithmic Experience (AX)
• Shin, D. (2020). Beyond user experience: What constitutes algorithmic experiences? International Journal of Information Management, 52, 102061.
• Shin, D., & Park, Y. J. (2020). Role of fairness, accountability, and transparency in algorithmic affordance. International Journal of Information Management, 57, 102142.
• Zarouali, B., Dobber, T., De Pauw, G., & de Vreese, C. (2020). To be or not to be algorithm aware: A question of a new digital divide? Journal of Computer-Mediated Communication, 25(6), 422–442.
• Kosinski, M. (2024, October 29). What is black box artificial intelligence (AI)? IBM. https://www.ibm.com/think/topics/black-box-ai

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI