Why is creating a daily "support or oppose AI" survey necessary or particularly helpful to achieving the goal of demonstrating the dashboard technology?
During the epidemic, I published a manifesto calling on social scientists like me to make their data and analysis free and open, preferably in the form of online dashboards. Since then, it has been difficult for me to live up to my own ideals, much less prod others to take these costly, unconventional actions. Here I propose a demonstration project that will test my ideas for automating social science data collection, analysis and publication.
Specifically, I will write open-source code to link data collection, statistical analysis and visualization. I will leverage the tools and platforms that are familiar to social scientists. The goal is persistent, consistent, precise estimation of public opinion. One achieves this through repeated surveys. Repeated surveys are currently painful to administer, and thus are rare. This project aims to remove pain, bit by bit.
The ideal endpoint is that social scientists compose online surveys in Qualtrics - a tool many already use. Then, with a tiny bit more work and lots of computer magic, a daily-updated, publicly and freely available web dashboard emerges.
My code will provide automated respondent recruitment, data download, data analysis and visualization. Respondents will be provided by Prolific Academic. That is again a tool many are used to; it provides an API and a large subject pool with minimal overhead cost. I will write Python code to download data from Qualtrics and post it in convenient form (csv file) to the Open Science Foundation. The OSF provides an API and free storage for open science data. Data analysis will be conducted in R. Summary statistics and item-to-item comparisons are easily automable. I will also write visualization code in R using the tidyverse. Finally, Python code will write HTML files presenting the results.
The result should be an online dashboard that looks something like this: https://jasonjones.ninja/jones-skiena-public-opinion-of-ai/It will be automated and update daily. I will develop the process in an iterative, incremental way. Modular tools will take over the most automatable parts of the process, bit by bit.
I will demonstrate how to use the tools by repeating surveys I have run in the past. These are focused on US public opinion toward Artificial Intelligence. For an example, see https://www.theseedsofscience.org/2024-attitudes-toward-artificial-general-intelligence
The budget is needed for survey respondent recruitment.
Cost Item
$3,650 Daily "support or oppose AI" survey
$912 Prolific Fees
$4,562 Grand Total
I am approaching 20 years of professional research experience, and it has taught me a lot. I have published peer-reviewed research articles, blog posts and dataisbeautiful graphs with millions of views.
$4,562
My dashboard scientist manifesto: https://jasonjones.ninja/dashboard-scientist-manifesto/dashboard-scientist-manifesto.html
90% I have the skills and this year I will have the time.
Jason
8 months ago
Why is creating a daily "support or oppose AI" survey necessary or particularly helpful to achieving the goal of demonstrating the dashboard technology?
Jason Jeffrey Jones
8 months ago
That is intended as a demonstration project. The tool makes dashboards, but when it is done, it is just another github repository. Running a daily survey (on a zeitgesty topic) would test the system and visibly show off what it can do. @Jason