“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Alastair Boyle, global client partner and head of strategy at Essence.
Although artificial intelligence (AI) can help advertisers be more compassionate, it isn’t without potential pitfalls.
I have previously argued that compassion is about understanding people, their situations and needs, and then taking action to alleviate those needs. Advertisers can now use contextual data to compile a much more detailed picture of each individual who sees the ads – and the context in which they are seeing them.
The more sophisticated ads are using contextual details such as location, time and weather to predict an individual's needs and offer a solution. While it’s not possible for a human advertiser to adapt every ad in real time, we can use AI to do it for us.
I’m working with the widely accepted definition of AI as a system which perceives its environment and takes actions that maximize its chance of success. Some advertising optimization systems would fall under this definition, although it’s fair to say that these are not systems that will worry any professional Go player. In fact, without the right guidance, AI isn’t very good at advertising either.
One of the potential pitfalls of using AI to communicate is the “uncanny valley” effect. Originating from the field of robotics, this hypothesis holds that human emotional response to robots becomes increasingly positive as robots appear more human, but suddenly dips when robots look almost-but-not-quite human, bearing an “uncanny” resemblance. The same hypothesis can also be used to explain human responses to AI behavior, as well as our reaction to badly targeted ads.
When an ad is completely irrelevant, we are unlikely to notice it, never mind respond to it. As ads become more relevant to us and our situation, we are more likely to respond positively. However, when we reach a point where ads are highly targeted in the wrong way, we may respond negatively.
We’ve all seen unfortunate pairings of advertising messages and website content, such as ads for knives next to articles about a stabbing. We’ve all been stalked around the internet for days by creepy ads for that pair of shoes we viewed online but bought in-store or that hotel we considered but decided not to book.
The systems behind these ads perceive their environment and attempt to maximize their chance of success by showing a relevant message. But they’re getting it wrong, and our reaction is often worse than if they weren’t targeted at all.
To get it right, advertisers need knowledge, empathy and compassion. They need knowledge about an individual and their situation, which they can use to deduce an individual’s need state. And using compassion, advertisers can then propose the appropriate product or service to meet that need.
An overabundance of data in digital advertising can mean that we’re too focused on the accumulation of knowledge while neglecting empathy and compassion. Much of the conversation around programmatic advertising is focused on collecting and applying more data across more sources, rather than using that data to improve the way we connect with our audience as human beings. If we want to get out of the uncanny valley, we need to focus on using what we know to craft messages that are relevant, interesting and useful for real people.
When we use AI to help us, we have the opportunity to apply this thinking to every ad impression, not just a campaign as a whole. This presents an opportunity to create valuable interactions with every consumer at an individual level. As we’ve seen, there’s also a risk of alienating individual consumers if we get it wrong.
Despite the risks, it’s important for us to continue to explore ways AI can help advertisers be more compassionate. If it can enable us to create and display ads that precisely align with the emotional and/or rational needs of our audience, we could have a world in which we see fewer, more effective ads.