What do we mean by ‘automation’?
Automation is not new and has transformed many organisations and industries. AI and ML bring a new lens to automation. This new technology not only offers ways to automate decisions and better replicate people but also enables scale previously not possible. For example, Netflix uses AI to recommend what people should watch next (Netflix 2019). People ‘could’ make these recommendations; however, it would be infeasible given the millions of Netflix users.
It is essential to understand the similarities and differences between traditional automation to AI and ML. When we have a better understanding of where different technologies stop and start, we can make informed decisions about when to use AI and ML or even, where people are still needed:
Automation is a type of software that follows pre-programmed rules.
Artificial Intelligence (AI) is software designed to simulate human thinking.
Machine Learning (ML) is a subset of AI that starts without knowledge and becomes intelligent.
Automation is capable of doing things automatically without human intervention. Automation is everywhere and involves machines or software performing repetitive tasks, typically at scale. Automation can use AI; however, the majority of automation utilises traditional software to move data from one place to another. The difference between AI and automation is that AI aims to simulate human thinking. Put another way; automation works with data — AI ‘understands’ data.
What do we mean by ‘Machine Learning’?
ML is a subset of AI that not only simulates human behaviour, but also learns from data. ML starts without knowledge and becomes intelligent by identifying patterns in data. The definitions between automation, AI and ML blur primarily because automation can involve AI, and AI encompasses ML. Many references to AI are specifically referencing types of ML technology. Regardless of definitions, AI and ML are solving problems that were not traditionally possible, because, in essence, they can make judgments.
Terminology Check: Algorithm vs Model
When you train an ‘algorithm’ with data, it becomes a ‘model’. The distinction is vital as an algorithm by itself is not particularly useful. Once we have a ‘model’, we can make predictions on new data.
Algorithm + Historical Data = Model
Model + New Data = Prediction
Interested in more?
Our full white paper, The robots are not coming for your job – Part 1: Hybrid Solutions is available here.