How Efficient is Generative AI? Behind-the-Scenes of Kixx’s AI Campaign: “The World Without Lubricants”

2024-09-10

Photo of Changbeom Lee on the left, with his name written in English and his job title on the right.

Ask a question, and within seconds, you’ll receive an answer. Point out a mistake, and it will correct itself immediately, rarely making the same error twice. Is this a smart, hardworking employee? No. This is generative AI (artificial intelligence).

AI is rapidly transforming our daily lives and ushering in a new era in countless areas. From reshaping day-to-day activities to revolutionizing entire industries, the world is witnessing a profound shift driven by AI. Creative industries were no exception in embracing this power, with AI-generated imagery quickly becoming commonplace among businesses. Advertising, too, has begun to leverage generative AI by slowly moving away from real-life filming or 2D/3D graphics and sourcing a large portion of its imagery with AI.

As AI capabilities expand, office workers like myself, familiar with the increasing impact that AI is playing across all fields, are left wondering about the limits of AI. Particularly, as a marketer, who must stay up to date on the latest trends, I have been curious whether our jobs could eventually be replaced by machines.

This year, Kixx launched a digital campaign, "The World Without Lubricants,“ created entirely with AI. From storyboarding to video production and background music, AI managed every major campaign task, earning the campaign significant industry attention for its “AI-made” label. While AI greatly reduced production time and provided fresh perspectives humans could not think of on their own, Kixx encountered unexpected challenges. In this behind-the-scenes look at “The World Without Lubricants” campaign, discover how it was created and the current capabilities of AI in the creative space.

Kixx unveiled its latest brand campaign, “The World Without Lubricants,” to emphasize the vital role of lubricants in modern society. The campaign leveraged AI to create all aspects of the video, including images, footage and background music. By contrasting a world with and without lubricants, the highlighted their often-overlooked necessity through striking visuals. Within just two weeks of its release, the campaign surpassed 2 million views on YouTube.

→ Learn more about “The World Without Lubricants”

Why Show a World Without Lubricants?

Although lubricants play a crucial role in society by ensuring smooth and efficient operations of all machinery and equipment, they often go unnoticed by the public, which is precisely where the campaign began. It aims to shift lubricants from their role as hidden champions to the spotlight. By creating such an extreme scenario, where lubricants suddenly vanish from the world, the campaign illustrates the devastation our world would face without them, emphasizing their necessity.

Without lubricants, machinery and all its components would overheat and break down, eventually leading to the collapse of social systems and the world as we know it. Unlike mainstream lubricant adverts featuring sleek cars driving through cities or deserts, this bold approach broke away from conventional advertising norms. AI played an important role in visualizing this dystopian world, enabling us to underscore the critical role of lubricants in a more impactful way.

Why Utilize AI as the Main Production Tool?

There were several reasons for choosing AI to create this unique vision. The tool offers unique perspectives beyond human imagination, further advancing Kixx’s forward-thinking brand value, while enhancing production efficiency. In fact, we focused on AI from the planning stage, because the value AI provides aligns with the value of lubricant since both contribute to a seamlessly functioning society. The pursuit of increased efficiency has been ongoing for centuries, and it was intriguing to learn that both lubricants and AI have played a central role in this development.

A shipwreck background with steam coming from the ship. White captions are overlayed discussing the importance of lubricants in machinery and industry.

Snapshot of AI explaining what it would be like without lubricants.

Not only this, but AI was also the most effective tool for visualizing a non-existent world, leveraging the latest technology. Traditional video production, whether live-action or 2D/3D computer graphics, is labor-intensive and time-consuming. In contrast, AI can generate images and videos within seconds or minutes based on prompts, enabling unlimited visual experimentation with relatively short turnaround times. For “The World Without Lubricants,” AI generated over 10,000 sources to create the film, a workload otherwise unimaginable with traditional methods, though the quality of the final product still depends on the prompts provided by humans.

Another reason we chose to use AI was to craft unique and eye-catching imagery. To boost Kixx’s global visibility, the video required striking visuals to complement the bold, destructive theme and generate virality. The role of AI was to go beyond traditional imagination, detailing and representing a world where everything collapses without lubricants. It created unexpected scenes based on a predictable scenario, such as abandoned cars along coastlines and discarded parts in the deep sea.

Images of rusting ships sinking in the deep sea and damaged vehicles along the coastline with a dark and stormy sky.

Scenes otherwise unable to visualize with traditional methods were successfully created using generative AI.

Damaged vehicles along the coastline with a dark and stormy sky.

An Unprecedented Industry Experiment: Challenges and Lessons Learned

Using AI proved to be a great asset to the production team, though it came with its own set of challenges to overcome. Contrary to expectations that a one-minute video could be made in hours just by inputting prompts, the process turned out to be more labor-intensive.

The first challenge was that AI lacks the power to “imagine” new things. Generative AI learns from existing images and creates new ones based on that knowledge, making it difficult to visualize things that don’t exist. Initially, AI produced shocking yet somewhat familiar images that didn’t capture the intended vision. Constant human intervention was necessary to reassemble and refine the generated images. The production procedure required inputting prompts like “depict an unimaginable dystopia” first, and then sifting through and recombining the generated sources to achieve meaningful results.

An AI-generated image of a cityscape through a window (left) and an AI-generated image of a futuristic cityscape surrounded by the sun and clouds (right).

Images – initially appearing familiar – needed to be recombined and regenerated to look more unique and striking.

An AI-generated image of a cityscape through a window.

An AI-generated image of a futuristic cityscape surrounded by the sun and clouds.

The second challenge was the inconsistency in the vast amount of data AI was learning from. At times, AI struggled to execute prompts or accurately interpret prompts it hadn’t encountered before. For example, a request for a scene of a rapidly moving and overheating car engine piston initially resulted in images of incorrect or incomplete components, or pistons exposed outside the vehicle.Similarly, generating images of exploding car engines required multiple revisions due to the AI’s reliance on outdated data about engine types. While recently made vehicles typically have front engines, the AI images showcased rear engines exploding or created vehicles with old-fashioned designs. The vast amount of historical data that AI was trained on sometimes led to distortions, as it applied outdated information to current contexts.

B-roll images of engine pistons and a classic car exploding.

B-roll footage of engine piston overheating and car explosion.

A classic car explodes in the city with pavement and car parts flying.

AI-generated images and videos have a unique problem where correcting minor distortions often requires recreating the entire source from scratch. This challenge was both a limitation and a unique situation enabled by AI.

“The World Without Lubricants” project gave us a chance to peek into the rapid advancements in AI technology within the advertising industry. The project demonstrated AI’s ability to produce every element of a video from prompts while underscoring its current limitations and the substantial amount of human intervention required for problem-solving.

As generative AI continues to handle tasks beyond human capability, it will become a valuable ally. However, achieving high-quality results will still depend on human direction and error correction. This brings us back to the question: “Can AI replace humans?” As you watch the “The World Without Lubricants” video, imagine the future of AI and its potential impact.