Few technologies have sparked as much interest and controversy as GPT, as the world grapples with the implications of generative AI across everything from art to education.
The creative industry has greeted this new era of machine learning with equal parts excitement and horror, thanks to the technology’s eerie ability to mimic aspects of creativity. But AI shouldn’t be thought of as a substitute for human creativity. I view it as an intelligent tool that can extend my abilities in the creative field.
Unlike regular software like Photoshop, working with AI-generative algorithms is like having an intelligent support system that can vastly speed up the process of generating and executing ideas.
For example, when looking for a very specific image online, it can take hours to find the right one. With AI-assisted image generation tools, you describe what you need and the process is instant. If you don’t like what it comes up with, you fine-tune your request to get a better response. This boosts creatives’ ability to communicate their ideas with clients and creative collaborators, such as illustrators and designers.
At We Are Social, we’re experimenting with ChatGPT which is a conversational version of GPT3, GPT3 being a language transformer model created by OpenAI that’s trained on billions of written text samples in a variety of styles. We’ve worked with it in the ideation process, copy writing and creating assets.
AI can enhance the creative process by offering a different point of view or a variety of angles on a brief. When brainstorming ideas, you can also embark on creative role-playing where the AI takes the role of the client.
For example, I asked ChatGPT to pretend it was a consumer electronics client reviewing an idea for a campaign that projects characteristics of a phone, like the camera shape, onto buildings in New York and London. In its long list of recommendations, the AI ‘client’ suggested first taking the cost and logistics of the project into account and considering how the initiative can fit into a larger integrated launch campaign.
It's also useful for copy writing. Maybe you need a series of social posts for a brand event. AI-generated suggestions for posts can help you get to the final product more quickly, subject to your (human) changes, re-writes and modifications.
I’ve found ChatGPT’s ability to enhance how we communicate and collaborate together the most useful aspect of the technology, and it extends far beyond the work environment. Together with my three-year-old daughter, Zoe, I recently got ChatGPT to generate a bedtime story for her that had Zoe as the protagonist and featured a unicorn cat.
For all the possibilities this technology presents, the dystopian implications cannot be dismissed. At We Are Social, we recently investigated the current discourse around AI in art, with artists now questioning their relevance and continued existence.
For our investigation, we used AI to generate artworks inspired by its understanding of the biggest news headlines for any given day. We named our research Social Diffusion, a reference to Stable Diffusion, the text-to-image algorithm that provides its visualization functionality.
We were blown away by the AI’s ability to create something that was contextualised to a news story using just the headline and to select a style of art that related to the story. For example, for the news headline: “Hurricane Ian continues to batter Central Florida as residents cope with record flooding”, the AI came up with the following prompt and generated an accompanying expressionistic image: “The image is of a large body of water with trees and houses partially submerged. There is debris floating in the water and people are standing on the shore looking at the damage. The image is in the style of Abstract Expressionism, utilizing expressive brushstrokes to convey the feeling of chaos and destruction caused by the hurricane. The image is large and chaotic, with bold colours and lines representing the destructive force of the storm.”
This was achieved through a chain of data exchanges between multiple layers of AI that are optimized to process a facet of the artwork creation. But as fascinating as this process was, it also showed us how integral human intervention is to it. In this installation, the creation of art was the subject of exploration. But even in this case, you still need a human mind to put things into motion and chain these machine learning models together.
While this technology raises serious ethical questions, it is possible to figure out a way to proceed that ensures human creative output is protected. Napster was hailed as the death knell for the music industry when it first emerged, but, instead of destroying it, it forced the industry to develop streaming technology, revolutionising it in the process.
At We Are Social, we will always work with human creatives, illustrators, designers, artists, writers, craftspeople. Human creativity is critical to us and to the creative industries. But right now every creative technologist out there is experimenting with these tools. Will there be AI-generated creative agencies in the future? Perhaps, but we won’t be one of them. We believe AI has the power to supercharge, not kill off, our creativity.
Manolis Perrakis is drector of innovation at We Are Social Singapore.