AI is now an integral part of our lives and as a society awash with data, we rely on AI to process increasingly vast amounts of information. AI helps us determine the best experience and message for us as we socialise, search, shop, travel, and watch. But how can AI and machine learning aid the effectiveness of marketing?
Firstly, we have AI-powered programmatic, which has been in place for some time. Machine learning AI algorithms can learn and adapt from the patterns they encounter, gather large amounts of data to segment and serve more relevant content in real-time. The faster and more accurately we can identify, categorise, and predict, the timelier and more persuasive we can make our communications.
Secondly, AI is powering a step-change in personalisation, with the act of anticipation and the ability to be hyper-personalised. By processing data from a profile, AI can accurately predict what a consumer wants before they consciously know it themselves. For businesses such as Amazon and ASOS, using behavioural data to inform business decisions has been a longstanding principle. Emerging AI software, such as hyper-personalisation, is taking this to the next level. Decisions are becoming informed by increasingly sophisticated learning with the only limitation being the speed and scale of capturing good data.
AI-led assistants like Alexa and Siri may not have become as indispensable as anticipated, but we are not far from our smart fridge making soft drinks recommendations. These touchpoints are always on and now playing a normalised role within the brand ecosystem. In a scalable way, AI is now a representation of the brand voice and experience.
AI: the challenges
The AI I have just described is optimised to solve a particular set of problems – AI cannot improvise. For example, AI recognises a picture of a cat but not what a cat is. Rotating the very same image of that cat and AI now decides that it is guacamole. If the data AI has learnt from is only slightly broken, the system will go haywire. In a situation where context matters, AI is no match for human intuition. Patterns are easily recognisable for AI but it fails to identify the underlying motivation.
There are also privacy challenges and AI ethics is becoming a real talking point. Indeed, AI can embed and make permanent our own biases. If your training set is bias, your algorithm will also be biased. The potential of intelligent, scalable targeting anonymised through predictive modelling is vital as we find our way in the post-privacy era.
From the perspective of an advertiser, the future of AI is mixed. AI will solve more sophisticated use cases such as mass personalisation – we will need to ensure we have the correct data inputs, rules to process that data at scale and connection to creative automation. However, a sharp lens should remain focused on ethics and compliance. Done correctly, AI could be a means to make marketing, if not more human, more humanistic.
Lindsey Jordan is head of media creativity at MediaCom.