Four traditional challenges in using AI and Machine Learning have been: 1) identifying processes where AI/ML can help. 2) Coding appropriate algorithms. 3) Finding appropriate training data, and 4) Learning all of the above steps. Recent developments in high-level AI/ML systems have dramatically simplified the second step of coding. Scikit-Learn was one of the first widely used AI/ML libraries. It was begun in 2007 as a Google “Summer of Code” project. By 2012 it was a stable library in use by Python programmers around the world, so think that we’ve had usable Scikit-Learn for less than a decade. Google released TensorFlow 1.o in 2015 and TensorFlow 2.o in 2018. Each library has captured more powerful algorithms in simplified procedures as the field more rapidly evolves. Tutorials for TensorFlow 2.0 began to appear in 2019.
Training data has been collected, organized in appropriate tables/dataframes, and made available for learning AI/ML, but finding your own relevant training data remains a challenge. Tutorials for learning AI/ML algorithms are widely available to address challenge #4, but the original challenge of identifying where AI/ML can help remains a creative challenge (the answer seems to be “everywhere!”). Additional tools like R-Studio, Jupyter Notebook, Keras, and Google Colab make the AI/ML algorithms even more available to thousands of interested students and professionals.