Marketers are grappling with AI’s role in advertising

Shutterstock/John B Hewitt

At this year’s Cannes Lions International Festival of Creativity – the advertising world’s equivalent of the Oscars – Saudi Arabia finally won its first Grand Prix. 

Ad agency Wunderman Thompson Riyadh took home the top prize in the Creative Commerce category for “The Subconscious Order” on HungerStation, the food-ordering app.

The advertising campaign uses eye-tracking technology and artificial intelligence (AI) to tell consumers scrolling the app what their subconscious wants them to eat.

The same work recently won a Grand Prix in the Mobile category of Dubai Lynx, an advertising festival which is the Mena equivalent of the Lions.

• How tech is transforming Saudi’s Hajj travel industry
• Adnoc and G42 plan IPO of AI joint venture  
• $2bn funding behind Dubai’s digital startup economy

When a user has been scrolling too long on HungerStation, The Subconscious Order offers the option of watching special menus roll by automatically.

Eye-tracking software watches where the user’s eyes are focused, and then the app tells them what they subconsciously crave, and offers an order chosen just for them. The app uses “smart AI” to do this.

New data directions

There are a lot of conversations going on at the moment about AI, and for those in the advertising world some of the most interesting threads of those conversations pertain to the data behind it.

For years, followers of advertising have been unable to avoid the hackneyed phrase, “data is the new oil”.

Social media platforms such as Twitter and Reddit seem to have woken up to the black gold they are sitting on.

Previously, the value of user data lay in allowing advertisers to target consumer segments and individuals who might be more receptive to their messaging. 

But now all those archived conversations – written by real people in real-world language – are finding a new use: as a tool for teaching AI how to listen and answer back.

AI needs an incredible amount of data to train it, and that data can include conversations that normal people – like you and I – have online. 

Companies, such as ChatGPT owner Open AI, use publicly available sources of text such as the whole of Wikipedia and millions of e-books to provide the equivalent of an immersive language course for their algorithms.

These virtual training camps for AI are called LLMs: large language models. Social platforms are an invaluable source of informal conversation for LLM data lakes. 

Twitter owner Elon Musk recently limited the number of tweets users of his platform can see each day. He said this was to prevent AI companies using Twitter’s accumulation of real conversations to train their algorithms.

And on June 14, thousands of Reddit discussion forums “went dark” to protest a new policy of charging third-party apps for access to the site’s data.

Musk tweeted that Twitter’s “drastic and immediate” restrictions were necessary “due to EXTREME [Musk’s caps] levels of data scraping”.

He continued: “Almost every company doing AI, from startups to some of the biggest corporations on Earth, was scraping vast amounts of data. It is rather galling to have to bring large numbers of servers online on an emergency basis just to facilitate some AI startup’s outrageous valuation.”

Customer concerns

Data conversations about AI are also conversations about the value of next-generation ad campaigns, and where that value comes from. This may cause regular users to stop and think a bit about the worth of their online words.

For some years people have voiced fears about the amount of data companies gather on us, the digital public. Before the conversation shifted to AI, advertisers were the most evident beneficiaries of this information.

That’s why we opt out of cross-app tracking. It’s why Google has promised to phase out cookies: so companies will track us to interrupt us with sales pitches less.

The latest season of dystopian semi-futurist Netflix series Black Mirror opened with an episode called “Joan is Awful”. The eponymous protagonist discovers a TV show is being made about her in real time by streaming service “Streamberry”.

A “quantum computer” does this with AI, using data gathered from Joan’s phone and Mexican-American actress Salma Hayek’s licensed image. 

The concept is chilling, even while we laugh along with it. But I’m sure there are advertisers out there rubbing their hands in anticipatory glee that this might become possible.

After all, that is pretty much the Holy Grail of media planning. Produce content so personal and compelling that consumers can’t help but watch it, then stick in ads to persuade them to buy stuff.

When all you have is a hammer, every problem looks like a nail, they say. The shiniest new hammer is AI, and technologists and marketers are finding new ways to smack some tacks into the crossbeam of advertising and consumer communications.

Some of what AI offers will be amazing. Some will be scary. Some will be, frankly, irrelevant, and some will change how brands talk to their customers.

We will have to see where the data behind this will come from, and if and how the value of that data will be transferred between stakeholders.

Austyn Allison is an editorial consultant and journalist who has covered Middle East advertising since 2007