Understanding Generative AI Capabilities in Everyday Image Editing Tasks
This work identifies gaps in AI-based image editing for everyday users, highlighting challenges in preserving identity and avoiding unwanted changes, which is incremental but provides concrete data for improving tools.
The study analyzed 83,000 real-world image editing requests from 2013-2025 to assess generative AI capabilities, finding that only about 33% of requests could be fulfilled by top AI editors like GPT-4o, with AI struggling more on precise, low-creativity tasks than open-ended ones.
Generative AI (GenAI) holds significant promise for automating everyday image editing tasks, especially following the recent release of GPT-4o on March 25, 2025. However, what subjects do people most often want edited? What kinds of editing actions do they want to perform (e.g., removing or stylizing the subject)? Do people prefer precise edits with predictable outcomes or highly creative ones? By understanding the characteristics of real-world requests and the corresponding edits made by freelance photo-editing wizards, can we draw lessons for improving AI-based editors and determine which types of requests can currently be handled successfully by AI editors? In this paper, we present a unique study addressing these questions by analyzing 83k requests from the past 12 years (2013-2025) on the Reddit community, which collected 305k PSR-wizard edits. According to human ratings, approximately only 33% of requests can be fulfilled by the best AI editors (including GPT-4o, Gemini-2.0-Flash, SeedEdit). Interestingly, AI editors perform worse on low-creativity requests that require precise editing than on more open-ended tasks. They often struggle to preserve the identity of people and animals, and frequently make non-requested touch-ups. On the other side of the table, VLM judges (e.g., o1) perform differently from human judges and may prefer AI edits more than human edits. Code and qualitative examples are available at: https://psrdataset.github.io