Google and Apple are building their ecosystems by limiting third-party access to data. And users are afraid of surveillance, but want to receive relevant content. Co-founder of the advertising platform Buzzoola Natalia Solodukhova and CTO of the project Anton Gavrish talk about what is happening in the advertising target market and what role AI plays in this.
The accuracy of ad personalization is growing every year, while privacy issues are becoming more acute. There are many stories on the Internet about “sinister targeting” – advertising content on websites and in social networks, which frighteningly accurately corresponded to the user’s requests or obsessively reminded him of unpleasant events. WIRED magazine recently wrote about a girl who had a failed wedding, but for several months she received targeted special offers for newlyweds, and online stores persistently reminded her of her anniversary.
Sponsored messages also change content and tone, adapting to the views of different audience segments. For example, the oil company Exxon shows liberals content that expresses its concern for the environment, while conservatives, on the contrary, issues a link to a petition against the regulation of the oil industry. There are other precedents when algorithms know more about a person than himself. For example, TikTok often offers content to users by anticipating and even predicting their preferences, such as sexual orientation. Historian and futurologist Yuval Noah Harari wrote about this – he suggested that at some stage, machine learning-based algorithms began to control our tastes and even our will.
However, the possibilities of an advertising target are often overestimated. For example, the assumptions about wiretapping, which our smartphones are conducting in the background, have not yet been confirmed, although some applications do collect audio data with the permission of users (only we give it unconsciously, mindlessly ticking the Terms & Conditions). In many cases, we build conspiracy theories around banal coincidences. So, in the early 2010s, the media spread a story about a teenage girl who hid her pregnancy from her parents. But a promotional mailing from Target supermarket betrayed her position – the company sent the girl discount coupons for clothes and furniture for babies, and the family learned about the secret pregnancy. Many facts in this story do not agree, and even data analysts question the story – trading networks regularly hold promotions for certain goods, and this is just such a case.