MOSCOW, February 26. Despite the fact that recently people are talking more and more about the use of artificial intelligence (AI), full trust in it is only just being formed; hardly anyone would dare entrust him with an entire advertising campaign — the risks are too high, says Victoria Tkachuk, head of mobile advertising development for Beeline adtech (edtech), the advertising technology division of Beeline Business.
She spoke about the risks, advantages and problems of using artificial intelligence in the preparation of mobile advertising campaigns.
Speaking about the difficulties, the interlocutor noted that AI is not a person, which means it may not accurately understand communication standards.
Some problems that may arise include: misunderstanding of context (AI may misinterpret context or meanings, which can lead to advertising that causes negative reactions among mass audiences), insufficient consideration of stereotypes (using AI in creating advertising creative requires checking for unwanted stereotypes), lack of cultural adaptation (AI may create content that is not culturally sensitive to the audience, which may result in inappropriate or offensive content).
However, according to Tkachuk, AI can bring significant benefits. For example, it can analyze the effectiveness of advertising content and optimize it based on advertising campaign data, which helps increase conversion. In addition, artificial intelligence can be used to create advertising messages based on given parameters and creative ideas. AI also helps in analyzing market trends and consumer behavior, which makes advertising more relevant.
Artificial intelligence can also be used to generate original texts that attract the attention of the audience and are consistent with the brand. Additionally, AI helps in creating visual elements such as logos, illustrations, videos and other graphic elements that can be used in promotional materials. And finally, AI can be used to test different creative options and optimize it based on audience reaction, Tkachuk explained.
She added that today Beeline adtech is developing an AI designer of advertising texts that will help write it from scratch. The user will be able to set an abstract task, for example: “Write a short advertisement about discounts on sets of stationery.” The designer will offer several options to choose from. However, given the shortcomings of AI mentioned above, it is important that the advertiser additionally checks the creatives created using AI.
«»Currently, Beeline.PRODVENIE has implemented a technology that highlights “forbidden” and “controversial” words in the text written by the user. Thus, when composing the text, the user will see red highlighting of words and hints about which words cannot be used in accordance with the law or rules of the advertising platform. The yellow highlight signals that it is better to replace the word with another one that is suitable in meaning. If the user decides to leave the controversial word, the system will tell you what needs to be added so that the advertising materials comply with the rules,» said Tkachuk, whose words were cited in the Beeline.
She clarified that the main difficulty in the data processing algorithm for AI is the huge number of goods and services, their combinations, objects, phenomena that simultaneously obey different laws and may contradict each other. There are many such situations, each requiring an individual approach. For example, AI does not take into account a number of language features when analyzing the generated request. Thus, he will perceive proverbs, sayings and dialect words literally.
For this reason, texts created by the designer must be tested by “natural intelligence”. At the same time, this allows developers to enrich the data, which allows them to adjust the model used by the AI.
“»This is everything on which we train the AI for a specific task. This data settles in its database and is used, among other things, when processing requests from others users. This improves the performance of the AI.
However, keep in mind that there may be an intellectual property or business information issue here if you try to use AI to solve some of your business problems. For example, ask the AI to write requirements for a new product that has not yet entered the market, or add a piece of code to your code base. Given the value of commercial ideas, in this case it is better to protect sensitive information by processing natural intelligence,» Tkachuk said.
In general, AI helps prepare advertising campaign, reduce time, find unexpected patterns in data, but it is still an imperfect technology and will not replace specialists in this industry today, the speaker summarized.
Advertising, VimpelCom PJSC, erid: 4CQwVszH9pWwJ4TsXo5