Which creative tasks should we give to AI?

Give AI research support, starting points and rapid prototyping, while people remain responsible for the idea, creative choices and finished work. Use its summaries to understand unfamiliar topics and its suggestions to explore possibilities, not as finished content.

Research and starting points

AI can help writers understand technical subjects and explore language before they develop finished content. At The Frameworks, a writer describes using conversational AI summaries to get to grips with topics such as agentic AI, reducing the time spent unravelling complex subject matter. The useful task is orientation: helping a writer understand a topic well enough to continue developing the work, rather than treating a summary as publication-ready copy.

The same writer describes AI-generated headlines as often average or clichéd, but still useful when a suggested verb opens up a different creative route. Give AI the exploratory task of producing possible language; leave the writer to recognise what is useful and develop it into something worth publishing. The Frameworks presents AI as a creative springboard, not a substitute for the judgment involved in writing.

Sources: How a Coffee Chain Helped Me Make Peace with AI

Human direction and selection

AI can accelerate prototypes while people imagine the experience, interrogate possibilities and choose what to develop. For G42, The Frameworks brought strategists, designers, writers and developers together to work out how AI could accelerate content creation. The team used AI to prototype ideas rapidly, then interrogated, validated and selected among them. The Frameworks attributes the project's pace to that combination of tools and cross-functional creative work, rather than asking AI to determine the experience.

The Frameworks also argues that access to generative AI does not produce quality output without professional expertise: marketers still need experiences aligned with brand values and customer needs. A practical division of responsibility is to let AI make possibilities quicker to explore while people retain creative direction and approval of finished work. The World Federation of Advertisers reported in 2026 that major brand owners were operationalising AI governance through human oversight, approved or prohibited tool lists, output review and approval, and central processes for introducing new tools.

Sources: We’re creating new experiences with AI – but we still need imagination, Introducing: Small AI for brands