Ten simple rules for navigating AI in science
Fig 2
Info-graphics summarising how to navigate the field of AI in the form of a tube (metro) map.
The main (blue) line denotes the rules (1-10) presented in this paper, while the other lines (green, pink, yellow, grey, red, brown and cyan) dive deeper into each of the guidelines by providing key terms associated with each rule. The aesthetic choices behind the info-graphics imply that we draw on metaphors to help make our rules memorable. For instance, we indicate that concrete tools might change (“under construction”). Similarly, to help readers avoid pitfalls, we highlight a few points where we encourage the reader to pause despite the urge to quickly apply AI to real-world data (“viewpoints”). Tracks bending in either direction indicate that some of the terms and concepts reappear in later or earlier rules. The rules in this paper are ordered according to the flow of scientific explorations: Framing the scientific problem and finding AI algorithms (Zone 2), coding (Zone 3), and testing and interpreting results (Zone 4). We hope it helps the reader to navigate on the journey from novice (Zone 1) to expert (Zone 5) in the topic. The silhouettes of the elephant and coffee mug are both taken from Wikimedia Commons and are in the public domain, while the remaining pictograms, including the hammer and pick, the parking sign and the aeroplane, are Unicode characters (U+2692, U+1F17F and U+2708).
