TermSight: Making Service Contracts Approachable

TL;DR

TermSight addresses the challenge of reading Terms of Service (ToS) by introducing a multi-level, AI-assisted reading interface that preserves the original binding text. The system uses Information Snippets, a Power Meter visualization, Summary Snippets, and Phrase Scope to guide readers from contract level to phrase level while linking to the source text. In a formative study and a user study with twenty participants, TermSight improved perceived readability and navigability of ToS, while comprehension and recall did not diminish, highlighting both benefits and limits of intelligent augmentation. The work discusses design implications, risks of dark patterns, and the need for policy-level solutions to reduce contract complexity.

Abstract

Legal contracts govern much of our society, but their specialized language is difficult for non-experts to read. While AI has enabled simplification of complex language, legal contracts pose unique challenges because of their connection to readers' values, ambiguity, and legally binding nature. Based on a formative study (N=20) using Terms of Service (ToS) as example contracts to study challenges in contract reading, we developed TermSight, an intelligent reading interface to probe the opportunities and challenges of designing augmentations for legal text. TermSight guides readers to relevant clauses with color-coded plain-language snippets of information and contextualizes ambiguous language with definitions and hypothetical scenarios. Importantly, TermSight's features always foreground the original, legally-binding contract text (e.g., linking to associated clauses). Our within-subjects study (N=20) demonstrated the opportunities of TermSight in making ToS significantly easier to read and navigate while revealing the challenges of augmenting service contracts such as ToS.

Figures (25)

• Figure 1: TermSight provides multi-level support for reading Terms of Service (ToS). At the contract level, TermSight visualizes the relevance and power balance of content in each policy (1). At the document level, TermSight chunks, summarizes, and categorizes content into one-sentence plain-language summaries (Summary Snippets) highlighted with colors that reflect power and relevance (2). Readers can click the Summary Snippet (2) or Highlight Bar (3) to navigate between the two. At the phrase level, TermSight offers phrase definitions and hypothetical scenarios for unfamiliar and ambiguous phrases (4).
• Figure 2: Design conceptualization of Power and Relevance. Power refers to the degree to which a snippet grants control to the service provider or the user. Relevance refers to whether or not the snippet is relevant to the user's persona.
• Figure 3: Power Meter visualizes the distribution of power and relevance of Information Snippets within a policy (a). On hover, a preview of Summary Snippets is shown (b) with options to view the referenced Information Snippets (c).
• Figure 4: Phrase Scope first identifies jargon or vague phrases in the contract (a). Clicking an identified phrase opens a tooltip with definition (b) and hypothetical scenario (c) while allowing users to ask clarification questions (d).
• Figure 5: A flowchart of the implementation for (1) obtaining a list of Summary Snippets each referenced to a span of the input text (i.e., Information Snippet) and (2) classifying Power and Relevance for each Information Snippet.