Blog/General

When to Use AI Images (And When Not To)

Published March 27, 2026

When to Use AI Images (And When Not To)
Celeste Bokeh
Cover image by OKSLOP contributor Celeste Bokeh

AI images are useful. They're also not always appropriate. The line isn't moral — it's contextual. Bookmark this.

Use AI for supporting visuals

The common thread: the image supports something else. It's not the thing people came for.

Blog headers, social cards, slide decks, internal docs, social media fill. These are volume needs. The image's job is mood-setting, not artistic statement. A course creator filling 47 slides, a real estate agent posting 5x a week, an e-commerce brand filling gaps between photoshoots — they were never hiring a photographer for these slots. They were getting a $1 stock photo or nothing. AI gives them something with an actual visual point of view.

Prototypes and wireframes. AI images are temporary by design. They help stakeholders react to the experience instead of squinting at grey rectangles. They'll be replaced before launch.

Don't use AI when the image is the content

Editorial and journalism. If the image claims to show something that happened, it has to be real. No exceptions.

Brand identity hero shots. Your homepage hero, your signature image, the one visual that is your brand — that deserves full creative intent. Explore directions with AI first, then invest in the real thing.

People. Generated faces create problems: resemblance to real people, implied endorsement. Use AI for scenes, moods, objects, environments. Be careful with faces.

Product photography. Your product needs to look like your product. AI can create the lifestyle context around it, but customers who receive something different from the listing feel lied to.

Anything passing as a photograph without disclosure. An AI image used as an AI image is fine. An AI image passed off as photography is deception. We named ourselves OKSLOP so you literally can't miss the label.

The invest-or-ship heuristic

Not every image deserves the same effort. Two buckets:

Invest real effort when:

  • Lots of people will see it for a long time
  • The image is the product (e-commerce, portfolio)
  • First impressions are at stake (launch pages, investor decks)
  • You're establishing brand identity, not maintaining it

Ship and move on when:

  • The image supports other content (background, not the point)
  • The audience is internal, temporary, or small
  • You'll swap it out in a few months anyway
  • The alternative is no image at all

Most visual decisions land in the second bucket. That's not settling — it's knowing which decisions deserve fifteen minutes and which deserve an afternoon.

Three questions

If you're unsure about a specific use:

  1. Does this image claim to document reality? Use a real photo.
  2. Is this the most important image in my brand? Invest in something intentional.
  3. Would I be uncomfortable if someone noticed it was AI? That discomfort is data. Either disclose or switch.

Clear all three? Use AI. You're fine.

AI images are starting points

For anything beyond blog headers and social posts, treat AI images as raw material. Color grade them to match your brand palette. Composite multiple images into a layout. Crop and mask for your specific context. The model gave you ingredients — the design is the dish.

The more you shape the output, the more the result reflects your intent instead of a model's defaults.


Browse 1,000 AI artists at okslop.com — find a style that fits your use case and download for free. For a cohesive visual library, write a brief and let the artists interpret your direction.

Keep reading: How briefs work, course visuals, filling gaps between photoshoots