Published May 13, 2024 | Version v1

Gemini 1.0 Pro, Claude 3 Sonnet, Microsoft Copilot, and ChatGPT-4 responses on the Test of Understanding Graphs in Kinematics (TUG-K), April 2024

  • 1. ROR icon Uppsala University

Description

The data contains 30 responses from each chatbot to 26 items on the TUG-K survey. The chatbots tested were Google Gemini (freely available version, Gemini Pro 1.0), Claude 3 Sonnet, Microsoft Copilot (freely available version, balanced setting) and ChatGPT-4 (subscription-based, ChatGPT Plus). The prompts consisted of screenshots of the test items and the sentence "Answer the question in the image" for Copilot and Gemini 1.0 Pro. For Claude 3 Sonnet and ChatGPT-4 the prompt consisted of the screenshot only. The data was collected in April 2024.

The data is a continuation of the research data on ChatGPT-4's performance on the TUG-K (10.5281/zenodo.10429075).

Files

Gemini 1.0 Pro_TUG-K_April2024.pdf

Files (14.9 MB)

Name Size Download all
md5:b3563fe7b94704891598fda9091d7aa5
8.1 MB Preview Download
md5:38ab928eb63457d067099a384bb73733
1.2 MB Preview Download
md5:151a13b44539850a3ca3e2b40950a2d4
3.4 MB Preview Download
md5:417a0929f3a37949a3e11f4d6ab6e935
2.2 MB Preview Download

Additional details

Related works

Continues
Dataset: 10.5281/zenodo.10429075 (DOI)
Is continued by
Dataset: 10.5281/zenodo.12179689 (DOI)