Real or fake?  Manipulative or justified?

  • A charity uses composite stories to reduce expenses
  • A charity uses composite stories to protect the anonymity of beneficiaries
  • A photographer edits photos to reduce sensationalism
  • A photographer edits photos to increase sensationalism
  • A disaster relief charity uses photos from a prior disaster given safety and ethical constraints using real-time photos
  • Feeding an AI generator sample, real images from a conflict and giving it strict guidance to create new, similar images that show resilience and autonomy of those in need

We’ve always edited and manipulated copy and photos for good and bad reasons.  AI image generation has simply lowered the barrier to entry and made it exponentially faster and better.

In a series of experiments researchers tested AI images in several conditions,

  • Control – didn’t tell the participant it was an AI image
  • Test – told it was an AI image with no rationale
  • Test – told it was an AI image with an ethical rationale (e.g. to protect anonymity of beneficiary)
  • Test – told it was an AI image with a pragmatic rationale (e.g. to reduce marketing costs)
  • Test  – the above conditions for disaster vs. non-disaster appeal

I’ll cut to the chase.  The best scenario is not using AI or using it and not divulging (you’ll need to decide the ethics of this).   AI image use and awareness of it lowers the capacity for empathy.  I’m less likely to feel for the beneficiary if I know the image isn’t real.

But if you are going to use an AI image and divulge then the ethical rationale is much better than the pragmatic one.  This will mitigate some but not all of the negative effects on empathy when using AI images.

Context matters.  If it’s a disaster appeal, using AI and an explanation that is part ethical, part practical (i.e. not feasible, unsafe) is no different than a real image on donation intent, attitudes toward the charity and attitudes towards the situation.