Advertisement and the March of the Bots

Modern Artificial Intelligence starts somewhere around this statement — “I had not realized… that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.” This is how Joseph Weizenbaum once described the apparent effect of one of the very first chat bots the world has seen, ELIZA, a natural language processing program that Joseph created 50 years ago. For those not familiar with Weizenbaum, he is one of the fathers of modern Artificial Intelligence.

Half a century later, while ELIZA certainly would never pass a Turing test, “she” is arguably still one of the more capable Artificial Intelligence conversation programs out there.

Why is ELIZA significant when looking ahead into 2017 and beyond? How is this relevant to advertising?

Picture this scenario — We use deep learning to programmatically craft the optimal message to specific groups of individuals, followed by more machine learning and automatic data models to plan and buy media exposure, such that we deliver the right message to the right audience at the right time in the right place — the full cycle. We use advanced models to determine profiles, which encapsulate behaviour, emotions, and recommended actions based on those characteristics. We establish when and where our audience is the most receptive to the tailored message we want to deliver. When the above is successfully executed, we’re looking at quite a powerful machine with which we can help brands in their quest to stay relevant every day.

This is not where we stop, though. It’s not the end game. At this point, we have merely succeeded in delivering a relevant message to a specific audience. As outrageous as it may sound, that’s quite an achievement, though, all things considered. However, it’s still just a message.

Over the past few years we have become accustomed to services like Siri and Google Now. In the last year or two Cortana, Alexa, and Ok Google entered the scene, and just the other day Facebook announced Jarvis. These are all prime examples of AI implementations. While not conversational, they are Question and Answer systems that intelligently respond to instructions and queries.

This is, I believe, where we should take note. Consider for a minute advertisement that responds, advertisement that is interactive, that listens, that understands. Not in the clunky, awkward HTML5-mini-app banner way, where the audience is imagined to possess an unearthly desire to play virtual mini golf on a 160x600 pixel strip, for a chance to part with some of their earnings. Rather, in a way where, for instance, a direct response ad can take bookings, answer queries pertinent to the offer, and suggest alternative routes or solutions if the ad itself deems the current audience non-receptive to the offer on display. Imagine an awareness banner that can sense the audience’ changing environment and physical circumstances, and subsequently adapt language and tone accordingly.

We can call them Intelligent Instructional Ads — advertisement that thinks.

It would be the first step towards an immersive online advertisement experience. It still would not constitute the end game, but we would have come further. We would be able to say that we have evolved with the times. That alone, would be fairly revolutionary, considering that right now many advertisers are still preoccupied with getting the right message out.

Enter ELIZA, and her cousins, nephews and grandchildren.

What would it mean if online advertisement not only could deliver the right message at the right time, respond to queries and offer alternative outcomes, but also could hold intelligent conversations? We would effectively have achieved what we’ve been trying to do since the birth of the so-called “rich banner” ad — real engagement through online advertisement. What we would want to do with such an engagement platform is a question for an advertiser’s imagination, but going all the way back to Weizenbaum — it’s a relatively simple implementation with extremely powerful effects.

If 2016 was the rebirth of the intelligent bots, 2017 is likely to explode in AI assistants, voice controlled helper apps, smart booking engines, virtual customer care, programmatic pizza parlours, and so forth.

As advertisers, while we continue our quest to refine mass personalisation and intelligent media buying, we will both develop and experience artificial intelligence upfront in online advertisement in the New Year. Whether we will reach the conversational engagement stage or settle for instructional ads is yet to be seen, but the coming year is the year of the bots, and we’re in for a ride.

Johan Reveillard

Head of TBWA’s Digital Arts Network Dublin
Technical Director TBWA\Dublin