Designers debate whether AI should decide or just build

Some designers argue that AI should handle only the building work, such as research, first drafts, and prototypes, because these tasks offer external checks that expose errors within minutes (TLDR Design). This camp maintains that humans must retain control over sensitive domains like financial choices or psychological advice, where decisions lack built-in validation and carry irreversible stakes. For these practitioners, the primary risk lies in delegating strategy to tools that cannot be held accountable for the outcomes of their suggestions.

Other designers advocate for using AI to actively shape strategy by generating multiple, materially different approaches to complex problems (Salvador Carranza from What AI makes PossibNow). Christian Terwiesch and colleagues found that applying different constraints—such as requiring one approach to work with existing staff while another asks the customer to participate differently—widens the collection of potential solutions. When designers treat the AI as a practical creative partner, they can inspect, test, and refine these ideas through iterative feedback loops rather than treating the agent as a final authority.

These two positions highlight a fundamental tension in modern design work. One side prioritizes the safety of human-led decision-making for high-stakes outcomes, while the other emphasizes the power of AI to expand the range of options before a human commits to a path. Both approaches agree that the designer must remain the final arbiter of what is useful, whether they use AI to validate execution or to challenge initial assumptions.