R

There are multiple IDEs for R. The most popular one is RStudio, as it is developed by a company that is very active in contributing to R (Hadley Wickham and others). RStudio is cross-platform and open-source. It comes in two types: RStudio Desktop is a regular desktop app, and RStudio Server is meant for running on a remote server, to which you can connect through your web browser. Interestingly enough, the desktop version is in fact just RStudio Server running locally, with an integrated web browser based on Google Chrome.

Even though RStudio is very popular, it is not the only IDE for R. RKWard is an alternative IDE, aimed at easier learning of R for people coming from other statistical software, such as SPSS. It includes menus for common statistical analysis algorithms, such as getting descriptive statistics, running statistical tests, and making plots. It also includes an editor for data tables. RKWard is an open-source native desktop app that is developed on Linux, but has also been ported to other platforms, including Windows.

Furthermore, there are cross-language GUIs/IDEs, such as Jupyter, whose name is a portmanteau of Julia, Python and R. Jupyter can run different language interpreters, what it calls kernels, for each script that is open, therefore scripts in multiple languages can be edited at the same time in the same interface.

One thing to keep in mind is that there is a distinction between a programming language, such as R, and its IDEs, such as RStudio. Which IDE you use is up to your own personal preference. Which IDE you used to develop code does not matter, because in the end you just have an R script that can be run on any of the IDEs, or directly from the command line. What does matter is that the script is written in the R language, as the users will need to have the R interpreter installed in order to run the script. Likewise, the packages that you use in your script are also important, as the users will need to install them before they can run your script. Therefore, whenever you refer to scripts and packages, for instance when writing a thesis, you need to specify what language the scripts are written in, and what packages (and ideally what versions) you used. You should not mention which IDE you used, as that is irrelevant to the readers. For example, you should not write “the scripts are written in RStudio”, but rather, you should write “the scripts are written in R, using package lattice version 0.22”. In fact, you should also cite the authors of the language and the packages you used. R includes a handy function citation() to help you cite:

citation()

## To cite R in publications use:
## 
##   R Core Team (2025). _R: A Language and Environment for Statistical
##   Computing_. R Foundation for Statistical Computing, Vienna, Austria.
##   <https://www.R-project.org/>.
## 
## A BibTeX entry for LaTeX users is
## 
##   @Manual{,
##     title = {R: A Language and Environment for Statistical Computing},
##     author = {{R Core Team}},
##     organization = {R Foundation for Statistical Computing},
##     address = {Vienna, Austria},
##     year = {2025},
##     url = {https://www.R-project.org/},
##   }
## 
## We have invested a lot of time and effort in creating R, please cite it
## when using it for data analysis. See also 'citation("pkgname")' for
## citing R packages.

It also works on packages:

citation("lattice")

## To cite package 'lattice' in publications use:
## 
##   Sarkar D (2008). _Lattice: Multivariate Data Visualization with R_.
##   Springer, New York. ISBN 978-0-387-75968-5,
##   <http://lmdvr.r-forge.r-project.org>.
## 
## A BibTeX entry for LaTeX users is
## 
##   @Book{,
##     title = {Lattice: Multivariate Data Visualization with R},
##     author = {Deepayan Sarkar},
##     year = {2008},
##     publisher = {Springer},
##     address = {New York},
##     isbn = {978-0-387-75968-5},
##     url = {http://lmdvr.r-forge.r-project.org},
##   }

There are more IDEs, such as R Commander, but we will not be talking about them within the scope of this course. Nevertheless, feel free to use whichever IDE is your favourite!

Next, let’s look a bit more in detail at the IDEs described above and their configuration options.

RStudio

In the virtual machines provided in the course, RStudio is already installed for you. In case you are working on your own computer and would like to know how to install R and RStudio, see the RStudio website.

If you are new to RStudio take a look at the following summary YouTube video about how to use RStudio: Intro to RStudio (6 min). In it you will learn how to navigate the RStudio environment and even run some code. This will be helpful for later in the course.

Additionally, the first time you open RStudio it’s helpful to change some global settings. To do this go to Tools → Global Options… in the top menu bar of RStudio. The following box should appear. Set the settings to match that of the screenshot, namely, Never save the workspace to .RData file on exit and uncheck the boxes under History and Restore .RData. The effect is that RStudio will stop prompting you on exit about whether you want to save the workspace and reload it on next start, which will save you time answering prompts and save time opening RStudio. Most importantly, it will make sure that restarting RStudio leaves you with a clean environment, equal to what another user of your code would start with, so you can properly test your code, and restart RStudio in case something goes wrong to get back to a clean state.

Global Options

To make a new script, click on File -> New File -> R Script, and a new editor pane will open, allowing you to write code. When you write a line, you can run each individual line that your text cursor is on by hitting Ctrl+Enter. To install packages, aside from using the R console below (install.packages() function), you can install them through the graphical Packages pane at the lower right corner of the screen, where you can search for and install packages and their updates.

Jupyter

Jupyter comes in two types as well: Jupyter Notebook and Jupyter Lab.

Jupyter Notebook is an older interface that is based on the notebook concept, where you combine text with code in a single (.ipynb) file. Jupyter Notebooks are extremely popular in the Python community, as they are easy for writing documentation and tutorials (vignettes). They are displayed automatically in GitLab and GitHub. However, Jupyter Notebooks are less popular in R, because RMarkdown provides a similar solution for R (and recently Quarto extends RMarkdown beyond R, therefore directly competing with Jupyter Notebooks). An issue with Jupyter Notebooks is that, due to mixing of text and code, you cannot run a notebook from the terminal, or in fact use any IDE other than Jupyter Notebook. They are also not well compatible with Git, as multiple developers cannot edit the same file without causing spurious conflicts, albeit there are workarounds.

Jupyter Lab is a newer interface that allows directly editing R and Python files, in addition to opening Jupyter Notebooks. Its advantage is that it can edit multiple scripts in multiple languages at the same time, but its disadvantage is that it cannot provide functionality that is specific to any one language, therefore it is relatively barebones.

Like RStudio Server, Jupyter is mainly designed to be run on a server. It can be run locally by launching a server and then connecting to it through a web browser. However, since Jupyter is written in Python, it requires specific Python setup to install. We will learn more about how to run Jupyter in the Python part of the course.

The fact that Jupyter is open source and designed to work as a server has created opportunities for Google to create something of a Google Docs version of Jupyter, that is called Google Collaboratory, or Google Colab for short. Google provides every user of Google Colab with computer resources to run Python or R. It is therefore the easiest way to run Python without changing anything on your own computer. If anything goes wrong, you can always create a new Colab notebook and start over.

First, go to the Google Colab website and click on + New notebook to create a Colab notebook. The default kernel type in Google Colab is Python, but you can change it to R by going to Runtime -> Change runtime type and selecting R. Click Connect at the top right to start the kernel, so that you can run code blocks.

Note that Google Colab has a special character, the exclamation point !, which allows you to run Bash commands within Python code blocks. This is very useful and important for installing Python packages, as they are typically installed using the pip command from Bash. You can try running a code block like this to install and load the rasterio package:

!pip install rasterio
import rasterio

Python

Aside from Jupyter, you can use other IDEs for Python. A simple to install IDE is called Spyder, as it is its own Python package. You can also use more complex IDEs for Python scripting, such as PyCharm and Microsoft Visual Studio Code, but they are more complicated to install.

You will learn more about Python IDEs later in the course. For now, we will stick with using Google Colab to run Python code.

In Python

A simple package (module) in Python is simply a directory with a Python file in it. In Colab, this code will create a directory and a Python file containing a Hello World example:

! mkdir -p MyPackage
! echo "def helloworld():" > MyPackage/helloworld.py
! echo "    print('hello world')" >> MyPackage/helloworld.py

Now we can directly make use of our new helloworld() function in Python:

import MyPackage.helloworld as mphw

mphw.helloworld()

In the import name, MyPackage is the name of the directory, and helloworld is the name of the file. We rename MyPackage.helloworld to mphw for brevity, but it’s optional.

Well, that was easy!

Though our package is technically not a complete package yet, as it can only be run if we copy the file to every code project that wants to use our function. A true package can be installed and then run from any place on your computer. Thankfully that’s also easy to do! All we need is to provide a bit more structure and metadata.

The metadata file in Python is called pyproject.toml, and it has to be in a subdirectory with the name of the package, as the name will appear in the list of packages. The actual functions need to be in a subdirectory of that subdirectory, with the name of the package, as the package will appear in Python. Typically, both of these names should be the same, otherwise it will result in user confusion. Lastly, if we want to be able to just import the package to get our functions, without having to specify the name of the file that holds the functions we want to import, we should call the script with our functions __init__.py.

In summary, the package structure looks like this:

MyPackage
├── MyPackage
│   └── __init__.py
└── pyproject.toml

Let’s make our files match this structure! First, make the subdirectory MyPackage/MyPackage, and then rename our helloworld.py to MyPackage/MyPackage/__init__.py:

! mkdir MyPackage/MyPackage
! mv MyPackage/helloworld.py MyPackage/MyPackage/__init__.py

Next, let’s add the project metadata. We only need to specify the name of the package and its version. To prevent confusion, use the same name as the name of the two directories.

! echo '[project]' > MyPackage/pyproject.toml
! echo 'name = "MyPackage"' >> MyPackage/pyproject.toml
! echo 'version = "0.1"' >> MyPackage/pyproject.toml

Bam, we now have a true package! We can install it with pip:

! pip install ./MyPackage

And now we can use our function from any directory simply as:

import MyPackage

MyPackage.helloworld()

In R

Packages in R are also nothing more than an organised collection of R scripts. Compared to Python, there are a few more files necessary to make an R package, but thankfully, we also have tools that help to create one!

Once again, let’s make a Hello World function and put it into a file:

echo 'HelloWorld <- function() print("Hello world")' > HelloWorld.r

Let’s make it into a package! We use the function package.skeleton() to autocreate a package structure:

package.skeleton(name = "MyPackage", code_files = "HelloWorld.r")

## Creating directories ...

## Creating DESCRIPTION ...

## Creating NAMESPACE ...

## Creating Read-and-delete-me ...

## Copying code files ...

## Making help files ...

## Done.

## Further steps are described in './MyPackage/Read-and-delete-me'.

Let’s follow the instructions to read the file MyPackage/Read-and-delete-me. And then delete it:

rm MyPackage/Read-and-delete-me

Let’s see what is the R package structure:

tree -n MyPackage

## MyPackage
## ├── DESCRIPTION
## ├── man
## │   ├── HelloWorld.Rd
## │   └── MyPackage-package.Rd
## ├── NAMESPACE
## └── R
##     └── HelloWorld.r
## 
## 3 directories, 5 files

It’s a little bit more complicated than Python, but not by much. The script files are stored in a directory called R (as R packages can also include C++ code etc.), and our script file is already there. Next we have the DESCRIPTION file, which, like the pyproject.toml file we saw for Python, includes metadata about the package. You can open it and edit it. Lastly, we have the NAMESPACE file, which states which functions will be accessible to package users (as opposed to only internal to the package), and a man directory, which includes the documentation manual entries. We have two entries: HelloWorld.Rd, which is the description of our HelloWorld function, and MyPackage-package.Rd, which is a description of the package itself. These documentation files are written in an R variant of LaTeX.

Let’s install our package! First we need to build it, from the terminal:

R CMD build MyPackage

## * checking for file ‘MyPackage/DESCRIPTION’ ... OK
## * preparing ‘MyPackage’:
## * checking DESCRIPTION meta-information ... OK
## * installing the package to process help pages
## * saving partial Rd database
## * checking for LF line-endings in source and make files and shell scripts
## * checking for empty or unneeded directories
## * building ‘MyPackage_1.0.tar.gz’

As you can see, the package is just a zip of our package files, plus some metadata cache. Now we can use R itself to install our new package:

install.packages("./MyPackage_1.0.tar.gz")

## Installing package into '/home/osboxes/R/x86_64-pc-linux-gnu-library/4.5'
## (as 'lib' is unspecified)

## inferring 'repos = NULL' from 'pkgs'

Perfect, we can immediately load it and use it:

library(MyPackage)

HelloWorld()

## [1] "Hello world"

As a bonus, we can also see the documentation we had in the man files:

?HelloWorld

That was also pretty easy!

To summarise, we have learned that packages in both Python and R are nothing more than scripts that contain functions, in a particular directory structure. If we start developing our projects with that structure in mind, we can make packages out of our scripts very easily!

Project structure

Making packages is a great way to stay organized, keep track of what you are doing and be able to use it quickly and properly at any time. Packages are:

• Easy to share with others
• Dependencies are automatically imported and functions are sourced (reduces the risk of having broken dependencies)
• Documentation is attached to the functions and cannot be lost (or forgotten)

For these reasons, if you build a package, next year you will still be able to run the functions you wrote yesterday. Which is often not the case for stand alone functions that are poorly documented and may depend on many other functions … that you cannot find any more. So to summarize, packages are not only a way to extend functionality, they are also the standard way to archive and save functions.

To make it easy for you to make a package out of your code, you need to stick with a project structure that packages have. Another benefit of maintaining a consistent project structure is that it will make it easier for you to switch from one project to the other and immediately understand how things work.

To practice keeping a consistent project structure, in this course we will be following the structure below:

Project Structure Schema

• A main script at the root of the project. This script performs step by step the different operations of your project. It is the only non-generic part of your project (it contains executable code outside functions, including paths, already set variables, etc.). The file extension of this file will depend on what language you are using for your project. We typically call these files main.r and main.py, but it could also be e.g. task1.r.
• As we will be working with multiple languages throughout this course we will keep things organized by placing the scripts into their respective language sub-directories (R/, Python/, and Bash/). These directories should contain the functions you have defined as part of your project. These functions should be as generic as possible and are sourced and called by the main script. The way this is done depends on the language used by the main script. For example, in R you would write source("R/myfunction.R"). Whereas in Python, you would use import Python.myfunction.
  • Each file in the R and Python directory should ideally consist of a single function with the same name as the file itself, to make it easy to find. In Python it is common to combine multiple functions in one file, because typically you refer to the file (module) name in addition to the function name (through import MyPackage), but it is still good practice to keep each function in its own file (so you don’t get confused where each function comes from even if you do from MyPackage import *).
• A data/ subdirectory: This directory contains data sets of the project. Since Git is not as efficient with non-text files, and GitLab has storage limits, you should only put small data sets in that directory (<2-3 MB). Typically, you do not include it in your git repository at all; rather, this directory is created from your main script, and is used to store data downloaded from the internet. It can be safely removed after the script is finished running.
• An output/ subdirectory (when applicable), where you place the final result of running your script. This should also not be tracked by git: your scripts create the output, so there is no need to store it.
• A README.md file should be included, this file should contain a description of your project, along with a description of what other packages your script needs to function correctly, and any other instructions needed to correctly run your code.
• Finally, you should include a LICENSE.txt file with the software licence which you would like your code to have.

# Team Teamname (John Doe and Jane Smith)
# January 2020
# Import packages
library(terra)
library(sf)
# Source functions
source('R/download_data.R')
source('R/function2.R')
# Use sourced function to download data to the directory "data"
download_data("data")
# Load datasets 
postboxes <- st_read('data/postbox_locations.gpkg')
# Then the actual commands

# Team Teamname (John Doe and Jane Smith)
# January 2020
# Import packages
import geopandas as gpd
import matplotlib.pyplot at plt

# Import functions
import Python.download as download

# Use imported function to download data to the directory "data"
download.download_data("data")

# Load datasets 
postboxes = gpd.read_file('data/postbox_locations.gpkg')
# Then the actual commands

Working directory, relative and absolute file paths

At the end of the following section you should be able to explain the difference between the following:

• relative path
• absolute path
• working directory,
• And the following special directories:
  • .
  • ..
  • / or the root directory

In the R and Python examples above we load the datasets by indicating the file location from the data directory "data/postbox_locations.gpkg". However, you may have many data folders on your computer, for all types of different projects. So how does the system know to look in the correct one? Moreover, if you share your script with a friend the location of their project and data folders will be different than that of your setup. It would be a nuisance if they had to change all references to these files in their script. To deal with these issues, we use relative file paths.

In relative file paths, we don’t include the location of the project (working) directory itself, these paths are relative to the working directory (the “Project_Structure” folder). In the example above, the relative file path for the post box locations file would be data/postbox_locations.gpkg, whereas the absolute file path would be "/home/osboxes/Geoscripting/Project_Structure/data/postbox_locations.gpkg". The absolute file path refers to a file from the root of the entire file system. On Linux (and other UNIX-like systems like macOS), absolute file paths always start with /.

So what is the working directory? By convention, the working directory is the same as the location of the script which you are working on. This means that you can simply assume that whoever runs your script, will run it from the directory that your script is located.

To refer to directories or files that are within the working directory, we simply use their names. So to refer to a directory called R in our working directory, we type R. To refer to a file or directory within another directory, we type the name of the directory, a slash, and then the name of the file/directory, for instance, R/function1.R.

If we want to refer to a file or directory that is above the indicated directory, we use the special directory ... For instance, if our main.R is not in our project root, but located in the sub-directory demo (therefore our working directory is demo), we would refer to our function1.R file as ../R/function1.R. Another special directory is ., which refers to the indicated directory itself.

When making Git repository, you want to make sure that all your code is portable and self-contained, i.e. you can run it from any computer, and ideally using any operating system. That means that as a rule of thumb you should always use relative file paths in your scripts.

Documentation

As we have seen in the package examples, documentation of code is extremely important to understand what your code is doing. It’s not only for others reading your code, but also for yourself, when you have to revisit code you haven’t touched in a while. If you follow the good scripting habits mentioned above, you will already have your code divided into reusable functions, that are in their own function files. Each function should have enough comments to understand each step the function is performing. Comments are any words after the # character, in either language. Same goes for the main script, it should also have each step described. But you don’t need to make it redundant: if it’s already clear from the function name what it does, it’s useless to repeat it in a comment. Rather, you would describe sections of code at once, e.g. what a loop does by itself, rather than its indvidual steps. In addition, if you use code from someone else, make sure to obey the license of the code you are using, and give credit to the original author using comments.

In addition to comments of your code, each function should also have a description and an explanation for all of its arguments. In both R and Python, there is a special way to add this information in comments next to the function definition, so that help files can be generated out of them automatically. The systems have different conventions, but effectively achieve the same purpose.

In Python, the system is called Docstrings. It consists of strings using triple quotes, that effectively acts as a multi-line comment. Here’s an example of a Python function documented with a Dosctring using the official reStructuredText (Sphinx) format:

def hello(name):
  """Greet a person.
  
  Give a customised "Hello" greeting for the input name.
  
  :param name: A string containing the name of the person to greet.
  :return: A greeting string.
  """
  return("Hello " + name)

The Docstring will be automatically printed when you run help(hello), and will be converted into a nice webpage using tools such as Sphinx.

In R, the system is called roxygen2 and is provided by the roxygen2 package. Here’s an equivalent example in R, using roxygen2:

#' Greet a person
#' 
#' Give a customised "Hello" greeting for the input name.
#' 
#' @param name A string containing the name of the person to greet.
#' @returns A greeting string.
hello <- function(name) {
  return(paste("Hello", name))
}

The roxygen2 package was created to make it easier to write documentation in R, so that documentation does not have to be separate from the code. The roxygenize() function generates the documentation files from the comments automatically, and then the documentation appears when running ?hello in our example.

Summary

Increasing your scripting/programming efficiency goes through adopting good scripting habits. Following the guidelines above will ensure that your work:

• Can be understood and used by others.
• Can be understood and reused by you in the future.
• Can be debugged with minimal effort.
• Can be re-used across different projects.
• Is easily accessible by others.

The summary of good practice guidelines below is not exhaustive, but already constitutes a good basis that will help you getting more efficient now and in the future when working on scripting projects:

• Comment your code.
• Write functions for code you need more than once:
• Document your functions.
• Keep consistent style.
• Make your own packages, or at least keep a similar directory structure across your projects.
• Use version control to develop/maintain your projects and packages.

Online resources

Great, now we know how to find help about specific commands! But how do we know how and what to write in the first place? Even the most experienced programmers run into these questions, so it’s important to know how to find answers to them.

There are many places where help can be found on the internet. So in case the documentation is not sufficient for what you are trying to achieve, a search engine like Google is your best friend. Most likely by searching the right key words relating to your problem, the search engine will direct you to online documentation, a tutorial, or to some discussions on Stack Exchange. It is quite likely that the problem you are trying to figure out has already been answered before, and using these resources you should be able to solve your particular problem as well. However, you need to be critical about the information you find on the internet, as it may refer to old versions of the software you are using, or it may provide a workaround but not a real solution to the problem. And, of course, some of the solutions may simply not work for you.

Question and answer forums

However, it may also happen that you discover a bug or something that you would qualify as abnormal behavior, or that you really have a question that no one has ever asked (corollary: has never been answered). In that case, you may submit a question to an appropriate Stack Exchange (e.g. Unix & Linux for Bash questions, or contact the author of the package you are using (often by filing an issue on the package’s GitHub page).

Stack Exchange has a few rules, and it’s important to respect them in order to ensure that:

• no one gets offended by your question,
• people who are able to answer the question are actually willing to do so,
• you get the best quality answer.

So, when posting to Stack Exchange:

• Be courteous.
• Provide a brief description of the problem and why you are trying to do that.
• Provide a reproducible example that illustrate the problem, reproducing the eventual error.
• Do not expect an immediate answer (although well presented questions often get answered fairly quickly).

Reproducible examples (reprex)

Indispensable when asking a question to the online community, being able to write a reproducible example has many advantages:

• It may ensure that when you present a problem, people are able to answer your question without guessing what you are trying to do.
• Reproducible examples are not only to ask questions; they may help you in your thinking, developing or debugging process when writing your own functions.
  • For instance, when developing a function to do a certain type of raster calculation, start by testing it on a small subset file, and not directly on your actual data that might be covering the whole world.

One could define a reproducible example by:

• A piece of code that can be executed by anyone who can run the programming language you are using, independently of the data present on their machine or any preloaded variables.
• The computation time should not exceed a few seconds and if the code automatically downloads data, the data volume should be as small as possible.

So basically, if you can quickly start a terminal on your neighbour’s computer while he is on a break, copy-paste the code without making any adjustments and see almost immediately what you want to demonstrate; congratulations, you have created a reproducible example.

Let’s illustrate this by an example.

I want to move all directories with Star Wars film subtitles to the directory ../starwars, but not move any of the Star Trek directories. Here is a piece of code that can recreate my directory structure:

mkdir -p films/{"the phantom menace","attack of the clones","revenge of the sith","a new hope","the empire strikes back","return of the jedi",\
"the motion picture","the wrath of khan","the search for spock","the voyage home","the final frontier","the undiscovered country","generations","first contact","insurrection","nemesis"} starwars
cd films

# I tried this, but it did not move the phantom menace, a new hope and the empire strikes back
mv *\ t* ../starwars

As you can see from this example, the problem is reproduced on any computer that is running Bash, and the changes are restricted to creating two directories, namely, films and starwars, which are easy to clean up afterwards.

Generative AI in coding

TODO: ADD MORE ABOUT COPILOT Another type of online resource that has recently been gaining in popularity is generative AI, such as ChatGPT. Generative AI models can be interacted with by asking it questions, including questions about programming. The AI responds by providing examples of code, explanations about what the code does, and how to run it. Of course, most AI solutions are not limited to code and will also answer questions on history, biology, quantum mechanics, and will even play Dungeons and Dragons with you, including throwing dice.

Generative AI models can be a great tool to enhance learning, as they can quickly answer specific questions and give coding suggestions. However, many of the limitations of web search apply to generative AI models as well (in fact, most of these models are something of a smart web search engine, as they are trained on a lot of text found on the internet). Therefore, you need to be very critical of AI-generated answers. The code that the AI generates may seem like it would solve your problem, but it may also do something incorrectly, such as calling functions that are no longer available, or even making them up altogether. Previously, many generative AI solutions were unable to provide references for their statements, and when asked provided a list of references and links that did not exist in reality, though this has improved in the past months. They may also answer questions completely wrong, but the explanation that they provide usually sounds quite convincing, therefore it may mislead you or make you second-guess yourself. When generative AI models are confronted about a wrong answer, they often insist that it is correct, and the longer you talk with a generative AI, the more it will get facts mixed up with its own previous answers, as it remembers and learns from its own output.

Generative AI tools can be chatbots, like ChatGPT, but they can also be tools that suggest code snippets as you write code, such as GitHub Copilot. The AI code suggestions are based on the same models and have the same pitfalls. But in addition, they may suggest code that was taken from software whose license is incompatible with the license of your own code, which could cause copyright issues. Some of the newer code suggestion models are able to provide references to where the code is sourced from, and the license it is under.

Some of the currently active generative AI tools are:

• ChatGPT - the original chatbot that started the generative AI trend. Made by a team of top AI researchers that formed into a company. The newest version, ChatGPT-4 is able to provide sources and references if you specify this in the prompt. It also allows the usage of custom GPTs, better suited to specific tasks, like the Python and R Wizard GPTs.
• Perplexity - an alternative chatbot built using ChatGPT-4o and Anthropic’s Claude 3.5 models that is able to provide references for its statements (and you can even pick which ones it uses to give you answers). However, it still gives biased output and may get confused with its own answers.
• Microsoft Copilot - Microsoft’s version of ChatGPT, also built on the ChatGPT-4 framework. It can also provide references for you if you ask. It generally allows you to access the newest version of ChatGPT for free, instead of having to pay for GPT Plus on the ChatGPT website.
• Google Gemini - Google’s version of ChatGPT. It has recently improved a lot compared to Google’s older models. Will also provide sources or references, if you ask for it. Can integrate with your Google Workspace (Gmail, Drive, YouTube), for example, it’s a great way to summarize a YouTube tutorial into bulletpoints or clear steps.
• Amazon Codewhisperer - code suggestion AI, free to use, but works only with some code editors.

Note that all of these generative AI tools are built on proprietary models, but there are open source alternatives such as Meta’s Llama 3.1 that you can use for your own applications.

Every day, more generative AI tools become available, increasingly embedded in the tools we use, like Google Gemini replacing ‘Ok Google’ on your phone and some web-browsers shipping with built-in models, like Brave browser shipping with Leo AI based on MistralAI’s Mixtral model. Despite the fact that generative AI is increasingly embedded, it remains important to check whether the results it provides are factual, accurate and if the output is compatible with the license of your own code.