Partakable Technology

Nardine OsmanIJCAI 2018 · DOI

TL;DR

This paper proposes a radical shift in technology development, empowering users to collectively define and control the rules governing their online interactions and data. By leveraging existing AI techniques like agreement technologies, natural language processing, and formal verification, the authors envision "partakable technologies" where users can actively shape their digital environment.

What Problem Does This Paper Address?

In today's digital landscape, users often find themselves at the mercy of centralized tech giants who dictate how their data is used and how they interact online. Recent scandals, like the Cambridge Analytica affair, highlight the profound impact this lack of control can have, even on democratic processes. This paper addresses the growing dissent against this power imbalance by proposing a framework where users, not just developers, have a say in the design and functionality of the technologies that mediate their lives. The core problem is how to bridge the gap between users' desire for control and the technical complexity of implementing those desires.

Methods

The proposed solution, "partakable technology," is built upon several key AI research areas:

  • Agreement Technologies: This forms the foundation for collaborative decision-making. Techniques like argumentation schemes and negotiation mechanisms are proposed to allow users to discuss and reach consensus on rules (norms) without needing technical expertise.
  • Natural Language Processing (NLP): To make these discussions accessible, norms will be expressed in natural language. NLP techniques are crucial for automatically translating these user-agreed-upon norms into a formal, executable language that the technology can understand and implement. The author is exploring methods to analyze verb usage and sentence structure to infer obligations, permissions, and conditions.
  • Normative Systems: These systems are used to enforce the agreed-upon norms. The paper draws inspiration from electronic institutions and lightweight coordination calculus, which provide frameworks for specifying and regulating interactions based on social norms.
  • Formal Verification: This is essential to ensure that the technology actually implements the norms users have agreed upon. The authors propose adapting model checking techniques, particularly those that allow for interaction-time verification, to confirm that the software behaves as intended.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Democratizing Technology Design: The central idea is to shift power from developers to users, enabling collective ownership and control over digital tools.
  • Bridging the User-Tech Divide: By using natural language for norm specification and agreement technologies for discussion, the proposal aims to make technology governance accessible to non-technical users.
  • The Power of Norms: The paper highlights how social norms, when formalized and enforced by technology, can effectively mediate complex online interactions.
  • Interaction-Time Verification: The emphasis on verifying system behavior as interactions occur is a critical insight for dynamic and evolving user-defined systems.

Key Results

This paper is a proposal and a roadmap, not an empirical study with quantitative results. The key "results" are the conceptual framework and the identification of existing AI techniques that can be integrated to achieve the vision of partakable technology. The author highlights their own contributions in areas like trust models for agent collaboration and dynamic verification of multi-agent systems as foundational to this proposal.

Limitations and Future Work

The authors acknowledge several limitations and areas for future work:

  • Integration Challenges: The primary challenge is integrating the diverse languages and mechanisms from different AI research fields into a cohesive system.
  • Language Ambiguity: While NLP is proposed, translating natural language into precise, executable code remains a significant hurdle.
  • Ethical and Legal Considerations: The paper stresses that a purely technological approach is insufficient. Collaboration with experts in philosophy and legal studies is crucial to understand and mitigate potential abuses and ethical implications of user-driven technology design.
  • Broader AI Integration: The authors suggest incorporating machine learning for norm synthesis and sentiment analysis for community feedback, as well as simulations for understanding norm consequences.
  • Multidisciplinary Collaboration: The success of this vision hinges on interdisciplinary teamwork, extending beyond AI to include social sciences and humanities.

Summary updated

BibTeX
@inproceedings{osman2018partakable,
  author = {Nardine Osman},
  title = {{Partakable Technology}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2018},
  doi = {10.24963/ijcai.2018/815},
}
Figure