By Pratibha Jain, Member of the Artificial Intelligence Society at King’s College London
The digital world today is ruled by the likes of Big Tech like Google, Amazon and Facebook. They run on data, the new oil, to fuel their revenue engines. Let’s understand how.
Hoarding user data today is equivalent to finding a treasure hunt full of jewels back in the day. With millions of daily users, tech giants are sitting on gold mines of data. They use this data to show us targeted ads in the name of “personalisation”.
Google’s ad revenue in the last quarter of 2021 was a whopping $61.2bn, while that of Meta was $32bn!
Indeed, if you’re not paying for a product, you are the product.
But, how do these companies make such accurate ads targeted to the right people at the right time? With the second jewel on their thrones: artificial intelligence.
Many businesses today rely on AI for targeting marketing and greater conversions. According to a Deloitte research report, 71% of AI adopters said that AI is very critical to their business.
AI and Advertising
Using Machine Learning, companies can first split their consumer base into segments, and show different ads to different cohorts, based on their needs and preferences. This ensures that the right ad for the right product reaches the right user. This also helps users as 56% of customers expect personalised recommendations from brands.
Natural Processing Language (NLP), a subfield of AI that analyses natural language and speech, can help refine marketing strategies for companies. One of the possible ways of doing this is using a feedback chatbot system where users type their feedback. The algorithm can then carry out sentiment analysis to tell companies whether their product or service has positive or negative attitudes associated with it. Then, these sentiments can drive the kind of ads that users will see.
But increasing concerns over data privacy often result in users not accepting cookies on websites. Cookies are strings of letters and numbers that websites save on your browsers when you visit them to enhance user experience.
To overcome this, a new type of advertising called “contextual advertising” was introduced. While behavioural advertising targets the users’ behaviour and preferences, here, the machine learning algorithm scans and analyses the content on the website, and places ads related to the content in the right places. This way, users are shown ads in their interest field, and don’t have to worry about their data being tracked.
Advertising and Social Media
In a world where 57.6% of the population uses social media, most companies turn to this medium for targeted advertising.
This strong web of social media keeps users hooked on the platforms, and by leveraging artificial intelligence algorithms to show targeted ads, users only find themselves entangled in the web. All of this may often result in polarisation, and breaches in privacy and ethics.
The Dark Side of Social Media Advertising
The algorithms of social media platforms like Facebook, Instagram, Twitter are trained to keep users hooked. The machine learning algorithms deployed by these platforms recommend content based on users’ current activity and preferences. While this might be good for the users as they get “personalised” content in their fields of interest, it often adds to human biases.
Reinforcing Biases
Confirmation bias is a human tendency to only process or accept information that they already believe. The algorithms play out perfectly to only show content that users believe and like. This puts them into “segments” of a particular belief, and thus narrows down users’ thinking, leaving very little scope for diverse ideas. Now, seeing ads only about things that interest users further reinforces their biases.
Social media algorithms often show “recommendations” based on the users’ peers’ content choices. This also includes political views of the peer, and may influence the political leanings of the user. In extreme cases, it may also lead to polarisation in society, causing political unrest.
Misuse of Data
Where there is data, there is always a concern over privacy and data sharing. Users expect companies to protect their data. However, companies often use users’ data to their advantage, which might at times be harmful for the user. For instance, if someone is addicted to gambling, the algorithm will only show them ads for gambling products, which is not only unethical but will also harm the user in the long run.
Misuse of data can also take unwanted forms of political influence. In 2018, Facebook exposed raw data of 87 million users to consulting firm Cambridge Analytica, which worked for the Trump Campaign during the US Elections. This data was then used for advertising during elections.
While Facebook was fined £500,000 for losing control over its user data, this incident sheds to light to how our data can be unethically misused without our consent, and influence us in ways we did not even know.
Fake News and Misinformation
Social media is the best place to cook up a fake story using wrong information. This misinformation acts as nothing but training data for the machine learning algorithms. Computer software follows a simple rule: Garbage In, Garbage Out (GIGO), that is, a wrong input is bound to produce a false output.
Biased data or incorrect data will make the algorithms biased, which will only get stronger with time. Moreover, AI systems could easily learn how to produce fake content via automated accounts and bots. How?
Many businesses use AI tools and automated accounts to create content on social media platforms for their accuracy and growth potential. But, if these accounts come across misinformation on the platform, they can also create content using wrong information.
After all, an AI algorithm learns from its environment.
So, AI can not only be trained on incorrect data, but also produce incorrect data! It’s a vicious cycle.
AI Advertising: A Bright Future?
All in all, if used ethically within the norms of data privacy policies, businesses can leverage AI for targeted advertising. However, they should ensure that the users have control of their data, the data is not misused and no ethical rules are breached.
Artificial intelligence has reached all walks of life. However, to be utilised in the best way for targeted advertising, it still has a long way to go. It is the key to unlock a great user experience for customers and also to help businesses grow, if used efficiently.