Introduction to algorithms, machine learning and AI
A view from Stephen Upstone

Introduction to algorithms, machine learning and AI

The difference between the latest terms being bandied around the industry.

As the field of artificial intelligence grows and enters the marketing mainstream it brings with it a host of new vocabulary and subtlety.

At Advertising Week, a panel on AI declared it and machine learning to be the same thing for the purposes of their discussion. But while the terms might be interchangeable from an event perspective, they equate to very different things.

As the fields of machine learning and AI become more entwined in our rhetoric it’s important we have a clear understanding of the difference – the last thing the industry needs is another "native", where definitions vary wildly from one provider to another.


An algorithm is a series of instructions written by a programmer for software to follow. It involves a series of steps to be followed verbatim. Algorithms are the building blocks for machine learning and AI.

Machine learning

Machine learning is a set of algorithms that enable the software to update and "learn" from previous outcomes without the need for programmer intervention. They are extremely important as it means delivery or targeting can be improved over time. For example, if a certain user responds well to one type of advert, that can then be drawn from in the future. Machine learning is not AI.

Artificial intelligence

AI is a whole new ball game. It is made up of a series of algorithms and includes machine learning. AI is designed to learn in the same way as a child. It sees the relationships between multiple factors for itself and builds assumptions based on those findings. This is called model-based learning, and allows AI to make far superior decisions than people because it can take many more factors, which may all be held in extremely delicate balance, into account, in milliseconds. AI incorporates algorithms and machine learning, and as such is a very different beast. 

If the above seems a little theoretical, a beautifully simple way to understand the difference between the three is to imagine baking a cake. An algorithm is the same as following a recipe created by a chef (programmer). You follow a set of instructions – mix the ingredients, heat the oven to 200c, and bake for 20 minutes. At the end of the process and you hopefully get a great cake.

But actually your oven is too hot and your cake burns. This is where machine learning would come in. It learns from the past that the oven gets too hot and so turns it down, improving your results.

AI is like a Michelin star chef. It knows the recipe, has grown up in the kitchen and learnt from thousands of past experiences. It also knows, in the past, raspberries worked really well with other chocolate recipes, and this is a chocolate cake, so why not add raspberries to this recipe? And maybe a splash of liqueur? By using AI you get something far greater than the sum of its parts.

AI is already delivering marketers exciting opportunities in 2017. It can seem impenetrable and as a technology it is brand new for many brands, agencies and tech providers in the ad space. But really it can be just as easy as baking a cake – for a Michelin star chef that is.

Stephen Upstone is chief executive and co-founder of LoopMe.