Understanding the role of technology in marketing
589 days ago
Stop talking AI and start talking machine learning
Whether it is Elon Musk calling for a ban on autonomous weapons, rumours of Tesco opening Amazon Go-style checkout-free stores, or Facebook abandoning a chatbot experiment after the bots started communicating in their own language, AI continues to be a hot tech topic in mainstream media.
AI also dominates the conversation in the marketing and ad tech industries, and is closely associated with programmatic, data processing and precision targeting, as well as semantic analysis techniques such as Natural Language Processing.
But what the marketing world often refers to as AI is actually a separate discipline known as machine learning. The two terms have become synonymous with one another, but it is important to recognise that they are not interchangeable.
Referring to technology as AI instead of machine learning, and vice-versa, is a misnomer, and marketers should start to understand the difference.
So what is the true relationship between machine learning and AI and where should marketers be putting their efforts?
The principles of AI revolve around the creation of technology that enables machines to function in an intelligent manner, and encompasses all devices that are capable of perceiving their environment and executing actions to achieve a specific purpose.
The term artificial intelligence refers to a computer's aptitude to autonomously carry out processes without programmed instructions.
Machine learning is a related - but separate - field altogether. Rather than creating devices that are capable of acting without instruction, machine learning focuses on achieving desired deliverables.
Described by Dr Marc Deisenroth of Imperial College London as the engine of modern AI, machine learning leverages algorithms which allow computers to learn from data. Rather than teaching computers everything they need to know through programming, they can be fed the data that allows them to learn for themselves.
Machine learning's application in adtech has evolved rapidly in recent years due to the massive amounts of online data now available to analyse. The analysis of this data allows for the construction of algorithms that can make data-driven predictions and decisions.
The more data that is available, the better these algorithms - and therefore the predictions and decisions - become.
Current applications of AI, such as IBM's Watson, use machine learning to find patterns buried in data, but technology - and the processors needed to power it - must advance well beyond this level to achieve true AI.
IBM is already developing a quantum computer to solve problems where there isnâ??t enough data to discover patterns, or where there are simply too many possibilities to process using classical computing. This will be the next phase in AI's progression.
When it comes to programmatic advertising, we are nowhere near the stage of super intelligence or even typical AI, but we are fully immersed in machine learning.
Often, though, it is the term "AI" that is more commonly - and mostly incorrectly - used by marketers.
The association with intelligent robots is something that captures the imagination, whereas machine learning doesn't sound quite so sexy. The word "algorithm" also has negative associations - it's a complex concept, difficult to explain in a short sentence, and means little to the person on the street.
However, the industry needs to take back control of the term and do a better job of explaining how machine learning is helping to develop algorithms, which make highly-targeted, exceptional advertising possible.
Machine learning is essential in optimising programmatic ad campaigns, using data insights to ensure the right message is delivered to the right user at the right time and in the right place.
It can identify which tactics and messages are producing the best results with specific audiences, so it can make ad execution continually more relevant and effective.
In the future, machine learning may also be able to capture reactions to ads, such as facial expressions and eye movements, and use this data to further enhance the ad experience.
There's no doubt that "AI" is an over-used term. Marketers must understand the differences between AI and machine learning and recognise at this stage they are tools that can be used to help them achieve their objectives, not words to be banded around as a means of over-inflating a company's tech expertise.
Taking the emphasis away from AI will allow marketers to concentrate on machine learning and begin to appreciate the vast capabilities of this technology in driving relevant, personalised ad campaigns, efficiently and at scale.