Skip to contents

reproducibleai

Description

{reproducibleai} is an R package of tools to help teams adopt a reproducible workflow for AI-assisted projects. This package contains a set of tools to assist data science professionals adopt efficient workflows during the early stages of AI adoption. We hold the opinionated view that the ethical use of AI requires data scientists to make well-informed, conscious decisions throughout the analysis process. These tools are a work in progress and designed to help our team experiment with developing new methodologies for AI-assisted workflows. These tools help to define a repeatable framework of chat instructions between sessions and across teams. Tools are provided to create a transparent audit trail to evaluate the use of AI.

Core ideas:

  • make chat sessions reproducible by recording the exact instruction modules used,
  • encourage reviewability (clear inputs/outputs and rationale),
  • support an auditable workflow as teams adopt AI tooling.

Authors

Installation

You can install the development version from GitHub with:

# install.packages("pak")
pak::pak("MVR-GIS/reproducibleai")

Quick start

  1. See which instruction modules are available
library(reproducibleai)
instructions_available()
#> [1] "chat-manual"    "goals"          "python-package" "quarto-book"    "r-package"      "shiny-golem"    "user-manual"
  1. Use a recommended recipe
recipes <- instructions_recipes()
recipes$r_package
#> [1] "chat-manual" "goals"       "r-package"
  1. Write instructions into your project
use_instructions(recipes$r_package)
# writes into dev/instructions/ by default

Vignettes 

# What vignettes exist for this package?
vignette(package = "reproducibleai")

Bug reports

Please open an issue at: https://github.com/MVR-GIS/reproducibleai/issues