The possibilities of artificial intelligence in advertising are often overestimated. Jan Suda, Media & Channels Strategist at McCann Prague, said this last week at the online conference Marketing Meeting: AI in Marketing, organised by Internet Info. According to him, AI can save time and make data processing more efficient, but it cannot fully replace human creativity.
The dream of all marketers and advertisers is for advertising to be produced quickly, cheaply and with high quality. AI is often presented as a technology that can handle all three parameters at once. However, according to Jan Suda, the reality is much more complex. In practice, time and cost savings range from zero to 40% and always depend on the specific activity. In many areas, AI does not bring any savings. "It is cheap, yes, but it depends on the case, and the magic is in the details," he added.
According to him, the key to effective AI deployment is a detailed audit of activities in a company or agency. For example, in the area of social media management, AI can help with part of the work, while strategic thinking, client work, and creative decision-making remain entirely up to humans.
Although AI can generate a large number of designs in a short period of time, according to Suda, it does not yet surpass human creativity in terms of quality. "Yes, it can generate much more work than a human in an hour, but it cannot yet be said that it does so with higher quality," he summarised.
AI handles simple or entertaining images quite well, but when it comes to realistic advertising visuals, it encounters limitations in detail, composition, and physical logic. Generating a pope in a feather coat is easy, but generating a person in a specific situation and activity is not so easy. AI outputs also require thorough checking and often post-production, especially for campaigns with high media budgets.
Similar limitations apply to texts and programming. In copywriting, it is necessary to contend with repetitive patterns, clumsy language and corporate jargon. "The rule of shit in, shit out applies – the quality of the input determines the quality of the output," Suda pointed out. "We expect AI to anticipate what we have in mind, but it doesn't understand that. We have to be great at inputting, and the prompts have to look beautiful."
According to Jan Suda, expectations of AI are enormous today, but at the same time often significantly detached from reality. Models are developing very quickly, but there is a gap between the ideas of the public, managers and the real functioning of the technology. CEOs, marketers and agency bosses would like to save on people and replace them with AI models. However, this is not happening on a mass scale, at least not yet.
AI is very smart, but it is essentially a blank slate. It often produces information that sounds convincing but is not true – it simply "talks nonsense". "If a colleague at work behaved like this, they would be fired after the fourth lie. The problem with AI, however, is that it is a complex and complicated discipline: it is very difficult to set up agent tools so that they do not make up information," he explained. Machines can mimic human thinking, but they lack empathy, emotion and real thought processes. So far, there is no real understanding behind them – only guesswork.
Where can AI help?
The greatest benefit of artificial intelligence today is in the area of processes and partial outputs, where a real shift towards greater efficiency can be seen. Jan Suda mentioned an example where a large company can use AI to offer a personalised approach to each customer, get closer to them individually and use technology to really help them and make them happy.
While marketers often overthink what is and isn't created using AI, the average consumer usually doesn't think about it at all.
Storyboarding has undergone a significant transformation: today, it is largely replaced by AI and is created photorealistically. On the other hand, drawn storyboards are more conducive to the creative process.
An important area is teaching AI models the visual style of the brand. The model learns the client's visual style, the style of photographs and specific rules for working with the brand. Clients can then generate outputs themselves based on prepared prompts. The key is, above all, the correct initial setup in the form of a clear definition of visual identity and a universal system that does not need to be constantly redesigned.
In the field of text, the role of AI as a brand copywriter is developing. The biggest challenge is to suppress the tendency of models to create generic, emotionless headlines. Extensive databases of examples, clear rules of "dos and don'ts," long-term fine-tuning, and consistent human control help.
AI is also gaining ground in research, for example in the form of synthetic focus groups and advertising testing. Newer GPT models are capable of providing structured feedback and are suitable for mini-brainstorming, verifying creative directions and preparing materials for discussion with clients. They can analyse target groups, perform segmentation and simulate a group that comments on creative proposals.
AI is beneficial in the area of social listening, where it saves time. It is recommended to connect to existing systems that already have data, with the integration of AI directly into these tools being a suitable solution.
Last but not least, AI streamlines campaign planning and evaluation, replacing a large amount of manual work.
"Within two years, most agencies will have tools that eliminate manual work. But even if AI did all the advertising, those without AI would be more successful because they would be different. I still think that the human element will lie in who controls the machine," concluded Jan Suda.
AI as a good junior
The Klik.cz group used artificial intelligence as a tool in the creation of a new television advertisement, which helped them to manage a demanding assignment within a limited time and budget, as described at the conference by Petr Princ, Head of ATL, ePojisteni.cz. The spot was created for the ePojisteni.cz brand, and its goal was not to build the brand but to elicit an immediate response. The campaign was intended to appeal to the existing target group of drivers and active travellers and to work in both Czechia and Slovakia with minimal adjustments.
The creators decided to build on the previously successful concept with a chimpanzee. They had approximately three months to prepare, with artificial intelligence playing the role of assistant in visualising ideas, creating storyboards and preparing production. However, the script and creative concept itself were created without the help of AI. "AI can't write a script for me yet, but even then it was already acting as a very capable assistant to the screenwriter," said Princ.
The use of AI also brought a number of limitations. For example, the technology was unable to generate Czech banknotes, consistently display characters in a single shot, or handle details such as the chimpanzee's teeth. The voiceover had to be provided by a live actor, and some elements had to be added to the spot in the traditional way. Nevertheless, according to Prince, AI significantly reduced the amount of work involved in the actual filming and post-production.
However, according to data from marketing mix modelling, the results of the campaign exceeded expectations. The new spot with the slogan "Compare smartly" achieved almost double the return on investment compared to previous campaigns and elicited a higher immediate response from viewers. It performed even better on the Slovak market.
"AI is not a creative genius, even today, but it is the best junior you have ever had. Just don't give it too much power," said Princ, summarising Klik.cz's experience with using artificial intelligence in advertising.
Tools used
Oldřiška Hradcová from Včeliště presented practical ways to simplify social media management using artificial intelligence at the conference. She emphasised that the basis for success remains regular communication and a well-thought-out publishing plan, the preparation of which can now be significantly accelerated by AI. She showed how to use tools such as ChatGPT to create topics, texts, publishing calendars and video scripts, and how to build a customised "project" for each client with brand context. She also focused on the current possibilities of AI in the field of visuals and videos, from Canva and Meta AI to tools for subtitling, voiceovers and music generation, which, according to her, enable smaller teams to achieve professional results with significant time savings. Specifically, she mentioned Captions for creating subtitles , Heygen, Veo3, Vertex AI, and EelevenLabs for creating voiceovers for promotional videos. Entire videos can be translated using Heygen or CapCat tools. Generating music for videos and advertisements using the Suno tool is very popular.
Source: mediaguru.cz