A simple guide to AI, Machine Learning and Deep Learning… or as simple as I can make it.

Before we become enslaved by our sentient robotic overlords, I thought I’d explain what Artificial Intelligence (AI), Machine Learning and Deep Learning are and how they impact on marketing and consumers.

Let start at the beginning; back in the 50’s we dreamt about robots making our tea (AI) and what we could do with our spare time, fuelled by the space race and science fiction it was a heady dream. Until that is the 80’s, where we started to ‘teach’ machines simple tasks (Machine Learning), the more times a machine completed a task the more it learnt and improved. We’ve now reached Deep Learning, where through the improvement of data storage, processing power and network speed algorithms can be used so that software can train itself through multiple iterations and vast amounts of data.

A simple summary of the three levels:

What is the impact?

True AI is still a dream, yes, we have robots that can run, win jeopardy but the amount of computing power required for the decision trees and the possibility of human brain is still a distance away.

Why is the term AI being banded around so much, why the hype?

Well ‘AI’ by its definition is the mimic of human behaviour. For example, Graymatter are a UK authorised partner of a software platform called Sentient Ascend. Can it predict the future, NO, can it make toast, NO, but what it can do is replace a human behaviour called A/B testing; which if you are not familiar with is used for developing web sites and user journeys by testing version A with one headline and version B with a different headline and analysing at the results.

This used to be a lengthy process of creating two versions, getting customers to use them, then reviewing the results. Then having to test two new A/B’s based on the outcomes and so on until you have an optimised version. Using a tool to replace this human behaviour, Ascend tests thousands of variations and combinations, measures the results and optimises it for you faster and with more combinations than any human could do in a given timeframe. Mimicking human behaviour this lives in the AI world. A simple add value – ‘do it faster that I can’ tool.

How is that different from Machine Learning? Machine Learning is formed from experience…

In the example above a machine matches using its learning tags to create a best guess based on the data it has, if this is correct, the bridge with sunset may become another tag.

Deep Learning, is a subset of Machine Learning with hidden layers (Neural networks) the software can train itself to understand its outputs but it also uses huge amounts of data.

Here is a simple example of a Deep Learning view:

Simple, right? No… here is an image of what a self-driving car sees whilst moving, learning constantly, the processing power required is immense.

So what do we use all this tech for? Imagine Deep Learning reviewing MRI scans or mammograms, the volume it could get through and the accuracy, or teaching children, learning and improving as it goes.

What is the impact on marketing of AI, Machine and Deep Learning?

What’s in store for us next? Tasks that a machine can do faster and better – brilliant! Embrace it and it will deliver a great ROI. Going all the way to driverless cars?

Finishing with a quote from Jason Pontin, MIT Technology Reviews editor-in-chief who states that:

Humans understand the tremendous potential in "embracing machines that are smart and powerful enough to become part of who we are." However, Pontin goes on to state that you must be aware of what's wanted from these smart machines, as "they are not solely loyal to our interests.”

Are we back to the beginning, awaiting the rise of the machines? No, although we are still far from that kind of computing power (Ubers flying taxi’s and computers that express art are a long way off) we can use AI to do things faster, better and more accurately today.

Gavin W H Anderson
Head of Strategy