BLUF

Artificial intelligence (AI) has been developed into much more than assisting in data entry; an AI can now learn (machine learning) and make decisions based on algorithms and feedback.

Summary

The term 'machine learning' was first used in the 1950s when Arthur Samuel stated that 'it gives computers the ability to learn without being explicitly programmed'. Artificial Intelligence (AI) can learn as follows:
  1. Supervised learning: is where an AI uses algorithms to learn from inputs and responses, and when additional inputs are applied, the AI can predict new results. 
  2. Unsupervised learning: is when an AI uses an algorithm to learn from examples without receiving confirmation. The AI can apply an algorithm to determine its data patterns. Historical data is needed for any of this learning to work. This data is required to train the AI, which can write new algorithms, create a model and predict an outcome.
  3. Reinforcement learning: is when the AI is provided with feedback on the consequences of previous decisions and applies this feedback in developing new solutions based on the new data. In effect, the AI is learning by trial and error.