Post by Roy Goodacre

I hope you like this GI-GO image. I remember learning all about Garbage In and Garbage Out (GI-GO) in the early 1990s when I got into artificial neural networks (ANNs). In the foundational work of Jure Zupan and Johann Gasteiger - who worked in the field of cheminformatics and artificial intelligence - in their book “Neural Networks for Chemists: An Introduction and Neural Networks in Chemistry and Drug Design” they demonstrated that no matter how complex or sophisticated an ANN is, its output fundamentally depends on the quality of its input training data. So, in this ‘modern’ age of AI and ML it doesn’t matter how much you polish a turd, if your data are 💩 your conclusions will also be garbage.