MAX SESSIONS THE INTERSECTION OF AI AND CREATIVITY My name's Mat Atkinson. I'm one of the co-founders of Ziflow, a collaboration platform that I'm proud to say helps creative teams deliver amazing creative work. Generative AI has exploded into public consciousness. It's arguably the most disruptive technological innovation since the creation of the internet. AI is impacting whole swathes of the economy, including and especially creative industries. So we're all asking with a mix of optimism and fear, will computers replace human creativity? Will AI and automation steal our jobs? At Ziflow, we think the answer to both questions is no. We're going to show you that by taking over the mundane tasks that are part of the creative process, AI is poised to become a powerful, creative collaboration partner. AI will help us achieve what Pablo Picasso spoke about when he famously said, "Art is the elimination of the unnecessary." AI has shifted firmly to the center of public awareness, but the media headlines about AI are often designed to stoke fear rather than optimism. In the case of people like us working in creative industries, these questions have been asked about us as well as by us. What we want to know is, how is AI going to affect me, my job, my career, my family, my kids? Are we going to be replaced by computers, or is the future more nuanced? Even exciting, will AI actually help us be more creative rather than putting us out of work? Is there art in artificial intelligence? Now, if we're going to talk about AI and creativity, it's worth asking, what is creativity? The concept of creativity is something that we, human beings, have an intrinsic understanding of. We can list great creative people, and we know great creativity when we see it. But when we get asked to define creativity, it's tricky. Neuroscientists say they've identified three distinct brain networks that are the key to creative thinking. The imagination network, which allows you to daydream, put yourself in someone else's shoes, generate mental imagery, and think about hypothetical situations. The executive control network, which is responsible for your cognitive control, attention and working memory, it helps you switch between different mental processes, especially important as creative thinking often involves shifting between various ideas and concepts. And the salience network, that helps you identify things you already know and link them to what you're seeing or thinking about in the present. So we could say that creativity is a skill that allows you to draw understanding of the world around you, connect those observations to your existing knowledge and imagine new applications of that knowledge in the world. Or, as Albert Einstein more succinctly put it, "Creativity is seeing what others see and thinking what no one else thought." So let's hold on to the ideas of having imagination, drawing on existing knowledge and focus or cognition. One of the misconceptions of creativity is the idea of the lone creative, sitting on their own creating something amazing. Think of the starving artist sitting alone in a garret in Paris, or Rodin sculpture of The Thinker. Vera John-Steiner, who was one of the world's leading scholars on creativity, said that this idea of the lone stoic thinker being the purest representation of human creativity was a romantic, but quite incorrect concept. The reality is that creativity comes from collaboration, joint thinking, passionate conversations, emotional connections and shared struggles. So back to the list of great creative people in history. Michelangelo collaborated with Sebastiano. Steve Jobs famously collaborated with Steve Wozniak. Pablo Picasso's closest collaborator was Georges Braque. And we all know that John Lennon wrote and composed with Paul McCartney. Even in ordinary day-to-day interactions, connections drive creativity. The day before I wrote this presentation, I had a quick Zoom call with Aaron, a very talented writer in our marketing team. In the course of a simple five-minute catch-up, we generated two hugely powerful ideas that will both help with enhancing our product and also provide great content for our blog. I wouldn't have had those ideas on my own, and neither would he. It was a connection that we created during that five-minute call that sparked those creative ideas. The reality for all of us is that we collaborate every day in our creative work. So let's also hold on to the idea that human creativity is fundamentally collaborative. Now, this isn't the time for a deep dive into AI tech, and I'm certainly not qualified to do that. But as many of us understand, AI at its core relies on neural networks and a rich source of data to process information and make decisions. By training on vast amounts of data, AI augmented systems generate new ideas and realistic content such as text, images, poetry, lyrics, music and more. To boil it down, generative AI has a much bigger repository of knowledge to draw on in the creative process than any one individual or even group could possibly have. And it can create a much higher number of cognitive links between data in that data set, almost instantly. This means that the creative content that AI generates can be more informed and possibly more sophisticated. But the question looms, is it more creative? To test just how creative AI can be, we asked ChatGPT to write a haiku about the moon. Haiku are hard to write as they're very short form, which is why they're seen as a great expression of creativity.
Here's what it came up with. SILENT ORB ALOFT, WHISPERING TO TIDES BELOW, NIGHT'S SOFT SILVER CLOTH. Not bad. In just a second or so, the algorithm was able to condense a huge amount of knowledge, concepts and ideas into this. The silver color of the moon, its shape, its effect on tides, the silence of nighttime and the way that the moonlight colors the landscape. I would have probably spent days, weeks, more likely, trying to produce something like this and still not have matched what we see here. So the question is, does that make the algorithm more creative? The answer is, not exactly. It's important to remember that what we see here is not based on original thought, but from the blocks of data sitting in the knowledge base of the AI. The AI can parse its vast data set and make connections using its algorithm, but it's not capable of original thinking. The starting point for this haiku was the question that I asked, the prompt, and what happened in practice is that ChatGPT didn't produce this poem immediately. This was the second of four versions that I asked it to create. I made the decision to request each new version and I made the decision to go back to the second version. So my human intuition was a critical part of the process, as was my human judgment about the preferred final version. What ChatGPT brought to the process was a huge database of references to the moon and the ability to connect the different moon-related concepts in that database. But what it needed was my initiation, my collaboration, and ultimately my decision making. We'll get back to the Art of the Prompt in a moment. But first, I thought it would be fun to ask ChatGPT, "How creative AI can really be?" Interesting answer.
I also asked ChatGPT, "How generative AI can help creative professionals create content?" Here's what ChatGPT had to say about that: idea generation, automating repetitive tasks, assisting in photo editing, brainstorming, testing, and optimization. So even the AI thinks that it cannot replicate human understanding and emotion and sees itself as a collaboration partner or copilot in the creative process. We agree with the AI. We see the powerful new AI tools that are available to us as collaboration partners. They reduce the time we need to spend on mundane tasks, and therefore give us more time to be creative. So the next question to consider is, given the availability of these new tools, what skills do we as creators need to master to get the best out of our new collaboration partner? Some key skills for you to think about are: understanding AI fundamentals, concepts like machine learning, neural networks and generated algorithms. We don't need to be experts, but having a basic understanding is important. Familiarity with AI tools, which tools fit into which part of our creative process. Understanding this in detail helps us pick the right tool for the right creative task. AI creativity, the Art of the Prompt, prompt engineering or even prompt as art. AI collaboration, working with AI on creative feedback and iteration, how to evaluate and refine AI-generated content to align with the creative vision. These are key skills we need to master to get the most from AI. So let's go deep into one of those skills, the Art of the Prompt. AI prompting skills are going to become high value, if not critical, for creatives. This will be a combination of prompt engineering and prompting as an art form. Here's a broad set of best practices for mastering the Art of the Prompt. Set clear objectives. Prior to typing in your first prompt, be absolutely certain about what it is that you're looking to achieve from the AI. Be specific. Don't paraphrase or self-edit. Good AI prompting isn't a time for brevity. You should give the algorithm as much information as you think it needs to achieve your goals. Provide contextual information. Don't always assume that the AI knows everything. Ensure it has the additional data points that it needs to come up with what you're looking for. Set creative constraints. You've heard the phrase "don't boil the ocean" and it holds true here. Don't expect the AI to deliver an entire campaign with only a few prompts. Specificity is important, so you should set boundaries around what you're looking for. Provide example outputs. Be specific about what it is you're looking for by giving the AI some examples. A well-known use case is using ChatGPT to create prompts for Midjourney. A crucial part of the process is training ChatGPT with how Midjourney's diffusion model works. By doing this, a single line instruction in ChatGPT results in a sophisticated prompt for Midjourney. And by pasting that prompt from ChatGPT into Midjourney, well, the results are pretty remarkable. Just as with human-to-human collaboration, human-AI collaboration is a deliberate process that can be learned and improved. Let's refer back to the haiku that I spoke about earlier. I utilized the Art of the Prompt in getting the initial response, but to me, it wasn't good enough. As I mentioned, this was the second of four responses that ChatGPT gave me. I regenerated the response three times and ended up choosing the second version. So not only did we start with my prompt, but the refinement process involved my feedback and ultimately, my human judgment on the best version. In other words, collaboration is key. The prompt is only your starting point. A virtual conversation with an AI is the way to get the outcome that you're looking for. Learning how to effectively collaborate with AI tools is a hugely important part of incorporating AI into our creative process. We need to ensure that our AI-generated content integrates seamlessly with our creative objectives. Once we have AI output, we need to think about content evaluation. Does it need the brief? How does it align with our project objectives, project vision and desired quality standard? Which elements resonate well with the project goals? We need to note inconsistencies, inaccuracies, or places where the AI output doesn't match our creative expectations. Feedback generation, we should compile constructive feedback based on that evaluation. We use the same criteria as we did for the original prompt. We're specific, we provide context and we reset constraints if necessary. Many tools have regeneration options, but the better our human feedback for the AI, the better the end result. Iterative generation, use our feedback to generate new AI outputs. This will guide the AI in generating improved content. Again, many AI tools have features that make this easy, but providing our human input at each stage rather than just relying on regeneration tools will give us better results faster. Comparison and analysis. We shouldn't forget to compare new AI outputs with previous ones. We should not assume that each iteration will be an improvement. AI regeneration can often run off in a single direction, so it's often necessary to come back to earlier iterations and reset. Team collaboration. We need to collaborate with our human team members during the feedback loop. We should gather input from team members to make sure that the feedback is well-rounded. We need to make it a collective effort, just as we do with human-generated creative. Feedback documentation. It's important to keep a record of our feedback and subsequent iterations of AI-generated content. This documentation can help us track progress, understand the evolution of our creative, and make more informed decisions. Learning, adaptation and best practice. Just as we're continuously learning from the AI output of each piece of creative, we need to continuously learn and improve our overall AI methodology. We should always be developing our prompting strategies, our evaluation criteria and feedback approach across all of our projects. In this way, we can create our own AI methodology and document our best practices, making the use of AI consistent across our teams. Ethical decisions and human oversight. Finally, as AI tools become universal in creative industries, several ethical considerations arise that we need to pay attention to. AI can't replicate the emotions and intent that human beings have. We need to provide oversight, address biases, and ensure that AI-driven tools haven't plagiarized or copied protected work. So as I close with what I hope was an insightful chat for you, let's see where we've been. We've explored AI's role in the creative process. We've seen how it's set to revolutionize creative industries, whilst, I hope, dispelling the notion that it replaces human creativity. We noted that the most creative minds in history often collaborated to achieve great things. So we should think of AI as a valuable collaboration partner, allowing us as humans to focus on genuine creativity. Picasso's wisdom, "Art is the elimination of the unnecessary," resonates as AI automates repetitive tasks. Collaboration takes center stage with AI as a collaborator, not a lone genius. Ideation, content creation, production and adaptation are all enhanced by AI. Ethical considerations demand human oversight, making us indispensable. So with all of that said, let's embrace AI as a creative ally, shaping a future where human imagination blends seamlessly with artificial intelligence. My name is Mat Atkinson, co-founder of Ziflow. Thank you for watching our webinar on "The Intersection of AI and Creativity."