3 minute read time.
IET Central London Evening Lecture Savoy Place, 13-Sep-2017
Speakers: Lisa de Bonis & Gary Jobe from Havas (a communication and marketing organisation)
Important note: These are the author's personal recollections and interpretations, which are likely to suffer from errors and selectivity. There is no endorsement from the IET or the speakers.


Lisa and Gary gave us a fast moving, informative and thought-provoking view of how AI (Artificial Intelligence or, more preferred, Augmented Intelligence) can be applied to several areas of marketing.

 

This is what I learnt... (with apologies if wrong)

 

AI is a natural evolution rather than a revolution; it's not about flying cars nor killer robots - Sci-fi writers and films have hyped it up on the years. Their job is to demystify and bring AI down to earth, to find practical solutions.

We already find AI in what we do every day, such as Google search, Facebook friend suggestions and Amazon purchasing recommendations. 

 

We can consider AI as the science of making machines more human, a science that has evolved since the 1950's Turing Machine. Underpinning AI is machine learning that takes actions based on data received (such as a Spam filter). Like a Russian doll, underneath is deep learning, working on lots of data.

 

AI has exploded now because we can access large amounts of data, have high power processors, parallel processing and "infinite" storage capacity. 

 

The attributes of a Cognitive System (like IBM's Watson) are the abilities to: Understand, reason, learn and interact.

A Cognitive system uses a large number of interconnecting API's for each of the many functions. With these, it can give insights into an individual's personality, split into a large number of attributes, such as sympathy, trust, melancholy. Gary showed this with a live demo analysing the text from a person's tweets.

Another example was an amusing demo of image recognition using Google Quick Draw to recognise Lisa's doodles. It continuously learns and improves, the more doodles it analyses - give it a go!

 

The three primary benefits can be summarised as:

  • Understand me as an individual; for a hyper-personalised engagement

  • Make every interaction matter; to be valuable and not get in the way

  • Collapse the customer journey; to enable purchasing decisions easily

 

The possibilities were shown in four interesting use cases:


  1. eagleAI (has landed…) which was developed for ITV news. During US election, it listened to what people were saying, fed on debates, speeches and social media chats, analysing the sentiment to predict the result. It got it right, confounding all the other predictions.

  • MS Sounds, developed to help Multiple Sclerosis patients to walk more and further. It collects multiple data to optimise the speed of moving and controls the pace through the rhythms of music.

  • SI, Sex Intelligence: A bot in Messenger to answer questions from teenagers who want more, good discreet information from a trusted brand. 'A robotic Sherpa to take them to the peak'.

  • The Next Rembrandt: Taught a computer to paint a new picture as if it was by the Old Master Rembrandt by studying his paintings, analysing the faces in a typical profile and the techniques he used. Using 3D printing to give paint depth, a very authentic and unique artwork was created. Using data to touch the human soul.

 

We saw a somewhat cheesy video from Amazon on their Echo Show hands free home assistant. This uses natural language programming with voice control to allow a family to order products, call their contacts, play music, etc - they only have to ask Alexa.

 

Wrapping up, no one knows what will happen next but for sure, success will come to those that are faster to adapt.

 

Lisa and Gary avowed an impressive ethical stance as they answered a range of very tough questions on:


  • The risk of thought control, influencing attitudes and opinions

  • Is AI being used for dumbed-down applications (e.g. Alexa ordering nappies)?

  • The intrusion into our lives - is it a high price to pay?

  • How much is our data worth? Does the data giver benefit as much (or enough) as the collector and user of the data?

Maybe the topics for a future discussion?