Generative AI tools such as Dall-E and ChatGPT are already affecting fields including graphic design, video and music. As these tools continue to evolve and improve, they raise considerations around copyright and attribution.
In some cases, generated content might not be directly copyrightable unless it involves substantial human input. Regulators and courts are also in the early stages of considering when and how intellectual property implications arise when AI is trained on copyrighted content to create new works.
C.J. Bangah, U.S. software principal and digital platform subsector leader at PwC, said AI has been in place in creative workflows in various ways for many years. However, the latest crop of generative AI tools is increasing awareness about how AI can be applied in creative workflows.
"Highly talented creatives worldwide can use AI to help bring their vision, at scale and with speed, to global markets without requiring massive upfront investments to translate an idea to action," Bangah said.
Narine Galstian, chief marketing officer at business and technology consulting service SADA, has found that generative AI tools have shifted her team's focus from the early stages of creating content to later stages by making the first draft faster and easier to whip up.
Increasingly, creative talent must be more involved in the middle and later stages of content production. Generative AI tools enable creatives to experiment with more choices in terms of creative direction. But in the middle stage, creatives need to verify that the output is what they intend. And at the end of the process, humans must edit and refine the content to ensure it's on brand and it accurately reflects the creator's aesthetic or intent.
How generative AI is affecting various creative fields
Alex Weishaupl, managing director at digital transformation consultancy Protiviti, said that -- far from earlier doom-and-gloom predictions -- the design and creative community is starting to enthusiastically embrace new generative services and tools.
Teams are finding interesting ways to integrate these tools into their workflows. In some cases, Weishaupl said, he sees a four- to fivefold increase in speed from the initial stages of idea generation and concept development to the final result.
Along the way, designers and creatives are taking advantage of generative AI integrations into tools to make mood boards, storyboards and visual inspirations, which enable teams to start exploring and iterating on interesting ideas much faster. Generative AI tools are also reducing the burdensome work of content operations for localization and personalization across local markets and smaller segments.
Nishant Jeyanth, practice director at research firm Everest Group, also sees generative AI tools playing a broader role in creative content lifecycles in enterprises. Tools such as ChatGPT and Dall-E are especially useful for crafting first drafts.
In graphic design, generative AI could automate repetitive tasks such as resizing and cropping, while ensuring consistency in design and quality. It could also make the processes of conceptualization, prototyping and design verification faster and more cost-effective. However, overreliance on AI could lead to issues such as a lack of human touch, loss of cultural context and aesthetic fatigue.
For video production, AI tools can provide automated transcripts and video tagging, as well as predictive editing and real-time feedback. Generative AI also shows promise for improving efficiency in postproduction and cross-platform optimization. But concerns remain around authenticity, trust, privacy, content pollution and loss of signature style.
In music, generative AI can produce adaptive compositions, synthesize and enhance voices, auto-tag, and generate metadata. AI tools can also offer users personalized song recommendations based on their mood. However, potential pitfalls include emotional deficiency, overemphasis on predictive analysis and difficulty measuring the quality of AI-generated music.
Generative AI raises new copyright and attribution issues
The creative industry, courts and regulators are still trying to figure out how generative AI will affect the future of copyright and attribution.
"No one has a definitive answer on how to handle copyright and attribution when the benefits of using large language models are so widespread and apparent," Galstian said.
Copyright and attribution remain gray areas for generative AI because, to date, there are few cases to show how courts will evaluate copyright claims. It helps that major generative AI providers are offering training data indemnity to soften these concerns. However, this will continue to be an area of uncertainty until legislators or judges clearly spell out how they will interpret such claims, Weishaupl said.
An additional complication is that copyright and business practices vary widely across industries. Written works, for example, have high standards and clear protocols for attribution and fair use. The music industry, in contrast, has a much more convoluted legal framework consisting of minute -- and sometimes inconsistent -- distinctions between copy and interpolation.
It's unclear exactly where generative AI creations will fall along this spectrum, including within the same domain and across different types of content. In Bangah's view, attribution specific to AI learning models will need to be considered from the perspective of human creators and AI learning agents.
"The monetization, access rights and responsibility of the enterprise in navigating the right way to harness AI without violating standards or expectations will be a continued area of focus in the future," she said.
The threat of automation in creative fields
Automated and augmented content generation tools could erase some jobs at the same time that they create others and make some existing work much easier.
Galstian said most enterprises will approach generative AI as a tool to enhance creativity, not replace it. Because today's crop of generative AI tools can create realistic yet ultimately fabricated content, rigorous evaluation and quality control will be critical aspects of generative AI-driven creative workflows.
John Singleton, co-founder of AI development platform Watchful, warned that automation could replace entire workflows if one worker is able to perform tasks that previously required a team of 10. Although this could let business users and some creatives invest more time into important work, many other creators could lose their jobs, creating friction alongside new opportunities.
Generative AI's potential impacts on the creative process
Marco Santos, Americas CEO at digital transformation consultancy GFT, said that at his firm, generative AI is already removing days and weeks of time that graphic designers previously used to find inspiration and best practices to inform their approach. Designers can now input a description of a general idea into an AI tool and eliminate the process of reviewing other platforms and sites one by one.
For instance, a designer might input a prompt similar to the following: "A minimalistic mobile banking interface that takes the user fluidly through a four-step process, showing no more than three fields per screen." The AI tool will then return several concepts based on best practices it identifies across the digital landscape, often within seconds.
Rather than using these outputs as the final design, designers can then use their own skills and creativity to review the elements across those concepts and conceive of something greater than the sum of its parts. This could include other internal considerations to which the AI doesn't have access.
Nisha Krishan, practice director at Everest Group, said that some of the various ways generative AI is affecting creative workflows in the enterprise include the following:
- Idea generation, such as AI-assisted brainstorming and content suggestions.
- User-centric content design, such as personalized content that is dynamically generated and displayed based on user behavior.
- Agile content production, such as automated content drafting based on content briefs and user insights.
- A/B testing and optimization, such as many variations of AI-powered A/B tests backed by user data.
- Trendspotting and predictive content, such as using AI trend prediction to proactively create content on emerging topics.
- Interactive and visual content generation, such as AI-generated quizzes, questions, enhanced visuals and other interactive elements.
- Content quality assurance, such as AI-enhanced proofreading and editing to maintain quality and accuracy.
- Data-backed insights, such as AI-generated insight reports that summarize user data and content performance matrices.
George Lawton is a journalist based in London. Over the last 30 years he has written more than 3,000 stories about computers, communications, knowledge management, business, health and other areas that interest him.