Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Taxonomy for Human-LLM Interaction Modes: An Initial Exploration

Jie Gao, Simret Araya Gebreegziabher, Kenny Tsu Wei Choo, Toby Jia-Jun Li, Simon Tangi Perrault, Thomas W. Malone

TL;DR

This paper identifies four phases of human-LLM interaction—planning, facilitating, iterating, and testing—and proposes a four-mode taxonomy (Mode 1: Standard Prompting, Mode 2: User Interface, Mode 3: Context-based, Mode 4: Agent Facilitator) to structure human-LLM collaboration. Through a two-stage systematic literature review of CHI/CSCW/UIST/IUI publications since 2021, the authors develop a primary taxonomy using a 5W1H-inspired framework (Who, What, When, How) and then systematically annotate and refine it across 73 papers. The resulting taxonomy captures a wide range of interaction patterns, including submodes such as UI for structured prompts, UI for varying output, explicit and implicit contexts, and team-focused agent facilitation, with examples drawn from contemporary LLM-enabled tools. The work provides a design-oriented lens to evaluate and compose human-LLM interactions beyond plain conversational prompting and points to future expansions across tasks and domains, including image/video generation and multi-venue literature reviews.

Abstract

With ChatGPT's release, conversational prompting has become the most popular form of human-LLM interaction. However, its effectiveness is limited for more complex tasks involving reasoning, creativity, and iteration. Through a systematic analysis of HCI papers published since 2021, we identified four key phases in the human-LLM interaction flow - planning, facilitating, iterating, and testing - to precisely understand the dynamics of this process. Additionally, we have developed a taxonomy of four primary interaction modes: Mode 1: Standard Prompting, Mode 2: User Interface, Mode 3: Context-based, and Mode 4: Agent Facilitator. This taxonomy was further enriched using the "5W1H" guideline method, which involved a detailed examination of definitions, participant roles (Who), the phases that happened (When), human objectives and LLM abilities (What), and the mechanics of each interaction mode (How). We anticipate this taxonomy will contribute to the future design and evaluation of human-LLM interaction.

A Taxonomy for Human-LLM Interaction Modes: An Initial Exploration

TL;DR

This paper identifies four phases of human-LLM interaction—planning, facilitating, iterating, and testing—and proposes a four-mode taxonomy (Mode 1: Standard Prompting, Mode 2: User Interface, Mode 3: Context-based, Mode 4: Agent Facilitator) to structure human-LLM collaboration. Through a two-stage systematic literature review of CHI/CSCW/UIST/IUI publications since 2021, the authors develop a primary taxonomy using a 5W1H-inspired framework (Who, What, When, How) and then systematically annotate and refine it across 73 papers. The resulting taxonomy captures a wide range of interaction patterns, including submodes such as UI for structured prompts, UI for varying output, explicit and implicit contexts, and team-focused agent facilitation, with examples drawn from contemporary LLM-enabled tools. The work provides a design-oriented lens to evaluate and compose human-LLM interactions beyond plain conversational prompting and points to future expansions across tasks and domains, including image/video generation and multi-venue literature reviews.

Abstract

With ChatGPT's release, conversational prompting has become the most popular form of human-LLM interaction. However, its effectiveness is limited for more complex tasks involving reasoning, creativity, and iteration. Through a systematic analysis of HCI papers published since 2021, we identified four key phases in the human-LLM interaction flow - planning, facilitating, iterating, and testing - to precisely understand the dynamics of this process. Additionally, we have developed a taxonomy of four primary interaction modes: Mode 1: Standard Prompting, Mode 2: User Interface, Mode 3: Context-based, and Mode 4: Agent Facilitator. This taxonomy was further enriched using the "5W1H" guideline method, which involved a detailed examination of definitions, participant roles (Who), the phases that happened (When), human objectives and LLM abilities (What), and the mechanics of each interaction mode (How). We anticipate this taxonomy will contribute to the future design and evaluation of human-LLM interaction.
Paper Structure (30 sections, 4 figures, 3 tables)

This paper contains 30 sections, 4 figures, 3 tables.

Table of Contents

  1. Introduction and Background
  2. Method
  3. Data Collection
  4. Stage 1: Initial Searching
  5. Stage 2: Paper Filtering
  6. Constructing the Taxonomy
  7. Stage 1: Development of A Primary Taxonomy
  8. Stage 2: Systematic Annotation of Each Paper and Refinement of the Primary Taxonomy
  9. Results
  10. RQ1: Four phases of a human-LLM interaction flow
  11. RQ2: Taxonomy
  12. Mode 1: Standard Prompting
  13. Mode 1.1. Text-based Conversational Prompting
  14. Mode 1.2. Text-based Conversational Prompting with Reasoning
  15. Mode 2: User Interface (UI)
  16. ...and 15 more sections

Figures (4)

  • Figure 1: Mode 1: Standard Prompting.
  • Figure 2: Mode 2: User Interface.
  • Figure 3: Mode 3: Context-based.
  • Figure 4: Mode 4: Agent Facilitator.