Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

8bit-GPT: Exploring Human-AI Interaction on Obsolete Macintosh Operating Systems

Hala Sheta

TL;DR

The paper addresses how rapid AI chatbot deployment can foster over-reliance and superficial engagement. It introduces 8bit-GPT, a Mac OS–based simulation that leverages slow technology and counterfunctional design to defamiliarize interaction with an LLM, creating friction that invites reflection on AI agency and anthropomorphism. The authors implement a complete artwork-driven pipeline using Basilisk II and Mini vMac emulators, a local-to-inference-server execution flow with Llama-2-13b-chatt, and a user study showing fair usability but meaningful reflective effects through interface friction and asymmetry. The findings demonstrate that nostalgic, laggy interfaces can provoke critical thinking about AI’s embeddedness and rhetoric, informing thoughtful, responsible HCI design and the broader social implications of human-AI interaction.

Abstract

The proliferation of assistive chatbots offering efficient, personalized communication has driven widespread over-reliance on them for decision-making, information-seeking and everyday tasks. This dependence was found to have adverse consequences on information retention as well as lead to superficial emotional attachment. As such, this work introduces 8bit-GPT; a language model simulated on a legacy Macintosh Operating System, to evoke reflection on the nature of Human-AI interaction and the consequences of anthropomorphic rhetoric. Drawing on reflective design principles such as slow-technology and counterfunctionality, this work aims to foreground the presence of chatbots as a tool by defamiliarizing the interface and prioritizing inefficient interaction, creating a friction between the familiar and not.

8bit-GPT: Exploring Human-AI Interaction on Obsolete Macintosh Operating Systems

TL;DR

The paper addresses how rapid AI chatbot deployment can foster over-reliance and superficial engagement. It introduces 8bit-GPT, a Mac OS–based simulation that leverages slow technology and counterfunctional design to defamiliarize interaction with an LLM, creating friction that invites reflection on AI agency and anthropomorphism. The authors implement a complete artwork-driven pipeline using Basilisk II and Mini vMac emulators, a local-to-inference-server execution flow with Llama-2-13b-chatt, and a user study showing fair usability but meaningful reflective effects through interface friction and asymmetry. The findings demonstrate that nostalgic, laggy interfaces can provoke critical thinking about AI’s embeddedness and rhetoric, informing thoughtful, responsible HCI design and the broader social implications of human-AI interaction.

Abstract

The proliferation of assistive chatbots offering efficient, personalized communication has driven widespread over-reliance on them for decision-making, information-seeking and everyday tasks. This dependence was found to have adverse consequences on information retention as well as lead to superficial emotional attachment. As such, this work introduces 8bit-GPT; a language model simulated on a legacy Macintosh Operating System, to evoke reflection on the nature of Human-AI interaction and the consequences of anthropomorphic rhetoric. Drawing on reflective design principles such as slow-technology and counterfunctionality, this work aims to foreground the presence of chatbots as a tool by defamiliarizing the interface and prioritizing inefficient interaction, creating a friction between the familiar and not.

Paper Structure

This paper contains 26 sections, 8 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Slow technology
  4. Counterfunctionality
  5. Methodology
  6. Running a legacy Macintosh computer
  7. Artwork concept
  8. Emulators
  9. Mini vMac.
  10. Basilisk II.
  11. Execution pipeline
  12. Local machine.
  13. Emulator program.
  14. User Study
  15. SUS Index
  16. ...and 11 more sections

Figures (8)

  • Figure 1: An example of a conversation on the main emulator program (P12).
  • Figure 1: Mean SUS index score across participant age groups.
  • Figure 2: The physical artwork installation. (a) From left to right: Two labeled floppy disks, a modern keyboard for interaction, Macintosh Plus monitor and keyboard, its manual, an Apple Mouse II, and a projector to display the emulator. (b) A rear view of the Macintosh Plus exposed case.
  • Figure 2: Mean SUS index score across participant occupations.
  • Figure 3: The main execution pipeline of the artwork.
  • ...and 3 more figures