Artificial Intelligence: Data, Creativity and Emotions
We live in an age where machines can pretend to be like humans. Artificial intelligence (AI) can learn and apply knowledge. The growing use of AI is shaking up the world of advertisement. Every day AI uses algorithms and analyses data to improve the speed and accuracy of predicting behaviours, targeting audiences and even creating advertisements without human involvement. But wait. It is not all it is cracked up to be. Let’s see what are the opportunities and challenges of using AI in the advertising sector
Before AI was easily at hand, advertisers had to adjust everything manually according to the budget (i.e. select the audience, create spreadsheets, look at the data and optimize the campaign by hand). Now, AI took the ‘wheel of optimisation’ into its ‘hands’ and made it more efficient. Automatic optimization of the ad spends, targeting and quick learning lead to better ad performance. Sophisticated and huge amounts of data are processed in seconds rather weeks. It allows to directly target people who would be interested in the product. The higher the relevance, the higher the conversion rate, the higher the revenue. Simple.
The obvious advantage of using AI in advertising is the fact that it is better quality than human-powered ads. As a result, forward-thinking marketers are choosing AI instead of people. Fair enough. The algorithm gets better as more data is fed to it. Think about the amounts of data that are collected by huge corporations (e.g. Facebook). You took a selfie, sent it to a friend — AI tracks your location. You talked to a friend about this incredible book that you would love to read — a book ad pops up on your screen.
The list of collected data could go on for pages. It could probably be made into a book (would you read it?). Although AI generates income, it raises concerns of privacy, accidents, discrimination and even manipulation of people’s decisions. Digital safety is at risk. How often do you read terms & conditions? Personally identifiable, sensitive information can be stored and used to manipulate you. A perfect example of this is the Cambridge Analytica case when personal data was used for a political campaign without users’ consent. As Julia Carrie Wong said: “data was amassed, data was extracted, and data was exploited”.
GDPR imposed restrictions for data collection. As a result, developing a GDPR-friendly AI has become a challenge. The primary purpose of AI is to learn from humongous quantities of data, but how can it learn, if the access to data is restricted? There are multiple ways AI developers are solving this problem, however, GANs (Generative Adversarial Networks) is the most interesting one. It uses two neural networks and less data more efficiently to achieve the same result. However, it does not provide a perfect solution as it still requires big data for training. For this day most of the methods imposed for AI development are limited. They may conform to one requirement but oppose the other. Future development should focus on developing AI that is compliant with GDPR.
But what about the creativity of AI? As of 2020, more than 30% of marketers are using AI to create next-generation ads. During the last few years, AI developers have put an acute focus on language learning and comprehension. Siri, Alexa and Google Assistant. This is the very beginning of a big change. Systems are programmed to learn natural language processing (NLP) and natural language generation (NLG) to create a great short copy generating revenue. For example, Phrasee writes email subject lines. Apparently, it’s better than the ones written by humans.
It is interesting to see how AI creations make sense. And don’t at the same time. AI can emulate the tone and style of writing, learn grammar and syntax. It can create nonsensical Harry Potter stories or generate beer names. Burger King even used it to narrate a video ad. It perfectly illustrates the public image of AI. Both powerful and stupid. However, AI is not able to produce higher-quality longer-form projects. And this is where human writers are superior. At least for now.
Future developments of AI will integrate neuroscience and quantum computing. Sounds promising, doesn’t it? We yet have to wait a few years for those developments. Although there will be a big leap in language learning and comprehension, humans remain the ‘ultimate software’ in ad creation for now. However, I believe that AI will never fully be a replace human creativity. AI is unable to emulate human emotions, therefore, not able to perfectly depict them in writing. When buying, we are led by our feelings rather than careful consideration. When we say “Hmm, I’ll choose this…” we mean that we like it and that’s why we’re doing it. When emotions regulate our decision making we choose very quickly. AI is determining actions on statistics, but not on emotions. Human behaviour is not something that can be perfectly put into numbers. People are irrational. And their behaviour is too.
According to Annette Zimmerman, it is just a matter of time until AI will be able to understand and simulate human emotion. It could be very useful for advertising purposes, shaping the direction of the ad and targeting people who feel in a certain way. (Maybe leaving the creation of the ad to humans *coughs*) But, wait a moment! Emotions are specifically private. Here we go again — breach of privacy. Before offering this development, companies should consider potential dangers, the accuracy and get users’ consent. If we are not thinking about the potential possibility of falling in love with your personal voice assistant, the future is looking bright.
Artificial Intelligence is changing the world. There is an endless pool of opportunities for the development of AI. Who knows, maybe soon enough many things will be automated. However, developers must take risks and challenges into account. Managing them and reducing the likelihood of potential pitfalls will optimize not only the industry of advertising, but also our life.
