MIT OpenCourseWare

Will AI Negate the Need for Science Photography?

The conversation between host Sarah Hansen and photographer Felice Frankel centers on whether artificial intelligence will render traditional science photography obsolete. Frankel recounts a collaboration with Nobel laureate Moungi Bawendi, where her artistic shot of nanocrystal vials was rejected in favor of a straight‑on documentary image for a journal cover, highlighting the tension between aesthetic choices and scientific documentation. Frankel then tests an AI model with a detailed prompt to recreate her nanocrystal image. The first attempt produces a cartoonish, inaccurate rendering, but a later run generates a strikingly realistic version, prompting her to admit that AI can soon match her technical skill. She stresses that AI‑generated visuals must be labeled as explanatory, not documentary, to preserve scientific integrity. Key moments include Frankel’s warning, “We must never ever say that a photograph is documentary if we used AI,” and the observation that the model substituted “quantum dots” for “nanocrystals,” showing AI’s evolving terminology. The discussion also notes that journals are beginning to draft policies on AI‑generated imagery, though standards for magazines and online platforms remain unclear. The episode underscores a looming shift: scientists and visual communicators will need clear guidelines distinguishing authentic data captures from AI‑crafted illustrations. Photographers may need to pivot toward curating and contextualizing images rather than solely producing them, ensuring transparency and trust in scientific communication.