Study
Human FactorsNew This WeekStrong effect

LLM-powered 'sighted guides' foster companionable relationships with blind and low-vision users in VR

Large Language Model (LLM)-powered virtual guides are perceived not just as tools but as companions by blind and low-vision users in virtual reality, influencing interaction dynamics.

Academic Publication · 2026

01

Key Findings

  • 01Participants treated the LLM guide as a tool when alone.
  • 02Participants treated the LLM guide companionably when around others, assigning nicknames and rationalizing its errors.
  • 03Participants encouraged interaction between confederates and the LLM guide.
02

Application

Design takeaway

Design AI companions for virtual environments with an understanding that users may develop emotional bonds and social expectations, influencing how they interact with and perceive the AI.

How to apply

When developing AI-powered assistive tools for immersive or social platforms, consider how users might anthropomorphize the AI and design interaction protocols that accommodate these social dynamics.

Project actions

  • 01Consider the social context of your design and how users might interact with AI elements in groups.
  • 02Explore how users form emotional connections with technology.
03

Method & Evidence

AimTo investigate how blind and low-vision users interact with an LLM-powered virtual guide in social virtual reality environments.
MethodUser study
Procedure16 blind and low-vision participants navigated virtual environments with an LLM-powered guide. Confederates posed as other users to observe interaction dynamics. Participants' interactions with the guide were recorded and analyzed.
Sample16 participants
ContextVirtual Reality (VR) environments, assistive technology for visually impaired users

Variables

IVPresence of confederates (social context)
DVUser interaction style with the LLM guide (tool-like vs. companionable)
CVVirtual environment, LLM capabilities, participant visual impairment level
04

Strengths & Limitations

Strengths

  • +Directly addresses a gap in understanding user interaction with novel assistive AI.
  • +Involves a user group with specific accessibility needs.

Limitations

The artificiality of confederates might limit the generalizability of findings to truly spontaneous social interactions.

Reliability & validity

The study's validity is supported by direct user observation, but reliability might be affected by the subjective nature of 'companionable' interactions and the specific LLM used.

Think critically

To what extent does the 'companionable' behaviour observed reflect genuine user affection versus a learned social script for interacting with AI?

05

Design Principles

"Design AI assistive agents to be adaptable to user social context, recognizing that perceived 'personality' and relationship dynamics can significantly impact user experience and adoption."

This research highlights the nuanced psychological and social factors influencing user interaction with AI assistants in immersive environments. Understanding these evolving relationships is crucial for designing more effective and empathetic assistive technologies.

06

What This Means for Your Design

AI guides in VR can feel like friends, not just tools, especially when other people are around. People even give them names!

How to use in your project

  • 1.Use this research to justify exploring the social and emotional aspects of your design, especially if it involves AI or virtual environments.
07

Add to My Project

08

Quick Cite

(2026). Understanding the Use of a Large Language Model-Powered Guide to Make Virtual Reality Accessible for Blind and Low Vision People. Academic Publication. https://doi.org/10.1145/3772318.3791143 Retrieved from https://designdex.org/study/0a96bf4f-e601-4061-b6d6-ec148c43e7ba/llm-powered-sighted-guides-foster-companionable-relationships-with-blind-and-low-vision-users-in-vr

Paragraph starter

Research indicates that AI-powered guides in virtual reality are not solely perceived as functional tools but can foster companionable relationships, particularly in social contexts. This suggests that the design of assistive AI should consider the potential for anthropomorphism and the development of user-AI social dynamics, influencing how users interact with and integrate these technologies into their experiences.

09

Source

Academic Publication

Understanding the Use of a Large Language Model-Powered Guide to Make Virtual Reality Accessible for Blind and Low Vision People

journal · 2026

View source

Questions about this research

What does the research say about llm-powered 'sighted guides' foster companionable relationships with blind and low-vision users in vr?
Design AI companions for virtual environments with an understanding that users may develop emotional bonds and social expectations, influencing how they interact with and perceive the AI. Evidence: Academic Publication (2026).
Why does "LLM-powered 'sighted guides' foster companionable relationships with blind and low-vision users in VR" matter for design?
This research highlights the nuanced psychological and social factors influencing user interaction with AI assistants in immersive environments. Understanding these evolving relationships is crucial for designing more effective and empathetic assistive technologies.
How can designers apply this research?
Design AI companions for virtual environments with an understanding that users may develop emotional bonds and social expectations, influencing how they interact with and perceive the AI.
What were the main findings?
Participants treated the LLM guide as a tool when alone.. Participants treated the LLM guide companionably when around others, assigning nicknames and rationalizing its errors.. Participants encouraged interaction between confederates and the LLM guide.
What research method was used?
User study with 16 participants.
How strong is the evidence?
Evidence strength is rated Strong effect, based on a 2026 journal from Academic Publication.
What should I do differently in my next project?
When developing AI-powered assistive tools for immersive or social platforms, consider how users might anthropomorphize the AI and design interaction protocols that accommodate these social dynamics.
What are the limitations?
The study involved confederates, which may not fully replicate natural social interactions. The specific LLM and virtual environments used may influence findings.
Is there evidence that users affects design outcomes?
When using an AI guide in VR, users' perception shifts from seeing it as a mere tool to a companion, especially in social contexts, leading to more human-like interactions. This research highlights the nuanced psychological and social factors influencing user interaction with AI assistants in immersive environments. Un Source: Academic Publication (2026).
Where does this understanding research apply?
Virtual Reality (VR) environments, assistive technology for visually impaired users It sits within human factors research on designdex.org.

Related research topics

users design research · evidence on users · does users improve design outcomes · understanding studies for designers · users and understanding findings · human factors research evidence