Where AI actually fits in design
AI has not replaced the design process; it has compressed specific parts of it. The biggest wins are in ideation (generating many directions fast), asset creation (images, icons, backgrounds), and repetitive editing (removing objects, extending images, tidying layouts). The parts AI does not do — understanding the user, defining the problem, exercising brand judgement, and making the final taste call — remain entirely human. The most productive designers treat these tools as a fast, tireless junior who can produce raw material on demand, while they stay firmly in the director’s chair.
Image generation: Midjourney, Firefly, and others
For generating original imagery, Midjourney leads on aesthetic quality and is excellent for mood, concept art, and striking illustration, though it lives outside most design apps. Adobe Firefly is the pragmatic choice for commercial design: it is trained for commercial safety, integrates directly into Photoshop and Illustrator, and powers features like Generative Fill for in-painting and extending images. DALL-E and Stable Diffusion round out the field, with Stable Diffusion offering the most control and customisation for those willing to run their own setup. The right pick depends on whether you value raw aesthetics (Midjourney), workflow integration and licensing safety (Firefly), or control and flexibility (Stable Diffusion).
In-app AI: Figma and Canva
The fastest value for most designers comes from AI built into tools they already use. Figma’s AI features help with UI work — generating first-draft layouts, renaming and organising layers, searching designs, and producing placeholder content — all inside the file you are already in. Canva’s AI suite (Magic Design, Magic Edit, background removal, and more) is aimed at speed for non-specialist and marketing design, letting you produce on-brand graphics quickly. The advantage of in-app AI is zero workflow friction: no exporting, no new file formats, no context switching. For day-to-day production work, that integration often matters more than having the single best standalone model.
Choosing the right tool for the job
There is no single best AI design tool; there is the right tool for each task. For original concept imagery and illustration, reach for a dedicated image generator. For editing and extending real assets inside your existing files, use Firefly’s Photoshop integration. For UI ideation and layout housekeeping, use Figma’s AI features. For fast on-brand marketing graphics, Canva. A practical stack is one in-app assistant for your primary tool plus one strong image generator, expanded only when a specific recurring need justifies another subscription.
Control, licensing, and the human layer
Two factors decide whether an AI tool is safe to use in real work. Control: prefer tools that let you guide output with references, style settings, and precise edits, rather than gambling on one-shot prompts — iteration is where production quality comes from. Licensing: confirm the commercial-use and training-data terms before shipping anything to a client or product, since they vary widely between tools. Above all, keep the human layer intact. AI generates options; the designer curates, refines, and decides. The tools that deliver the most value are the ones that make a skilled designer faster, not the ones that promise to design without one.