AI Boosts Early Design Work but Slows Experts During Execution
Generative AI accelerates brainstorming for designers but can slow experienced professionals once they move toward finishing work. A study of 192 students and 120 professionals found that the tool's impact depends entirely on which stage of the creative process a designer occupies.
Researchers at the University of Houston divided creative work into two phases: ideation, where designers generate many possibilities, and implementation, where they choose one direction and build it out. AI produced measurable gains in the first phase-raising novelty scores by 76%, relevance by 24%, and complexity by 97%.
The problem emerged during implementation. Expert designers spent 57% more time on projects when AI entered only at the finishing stage, even though their final creativity scores remained similar to those without the tool.
Why Experts Hit a Wall
Years of training create habits. When AI generated images using its own logic, veteran designers often stopped to revise, edit, and rebuild the output to match their practiced methods.
Screen recordings showed these professionals adding elements and editing existing ones far more frequently than less experienced designers. They were translating machine-generated work back into their own routines instead of accepting suggestions and moving forward.
Less experienced designers faced no such friction. They lacked the entrenched methods, so they could accept AI suggestions, borrow structure, and keep working toward a usable result.
Where AI Actually Helps Beginners
Among students without deep design training, AI improved novelty, relevance, and complexity during implementation. The tool handled parts of production they had not yet mastered, lowering barriers to finished work.
For beginners, AI was assistance. For veterans, it became an obstacle to overcome.
The Brainstorming Sweet Spot
AI clearly expanded the number of ideas people explored. Yet it did not trap them in endless indecision-most participants still carried roughly one option into the final stage, even after trying several machine-made possibilities.
Professionals reported more mental stimulation during brainstorming. Feelings of overload barely increased. That balance suggests early-stage experimentation opened the process instead of freezing it with too many options.
Building Better Tools
The next improvement should happen in how tools are designed, not just in the image generators themselves. Systems that adapt to users-letting beginners lean on automation while letting experts decide when assistance enters-would address the core problem.
The research came from poster design tasks, so results may not directly transfer to music, film, or architecture. Real projects also loop through many drafts rather than moving cleanly from ideation to implementation. Still, the pattern held across both students and working professionals.
For creatives looking to work effectively with these tools, the takeaway is straightforward: use AI to expand your options early. Once you're executing a finished piece, the tool works best if you control exactly when and how it contributes.
Learn more about how AI fits into your creative workflow with AI Design Courses and Generative Art Courses.
Your membership also unlocks:
