AI chatbots helped editor decode a faulty computer‑glasses prescription and get a replacement

David Gewirtz, a senior contributing editor, used three large language models — ChatGPT, Claude and Gemini — to decode a mismatched computer‑glasses prescription issued by his ophthalmologist and obtain a replacement pair that eliminated persistent monitor‑related eye strain. The episode matters because it shows how LLMs can serve as a practical cross‑check when written prescription notation conflicts with a patient’s stated working distances and lens requests.

Gewirtz visits the ophthalmologist regularly because of a family history of cataracts and glaucoma. The clinic’s workflow uses technicians to perform basic refraction testing while the doctor interprets results and conducts in‑depth eye health exams. Gewirtz is myopic and had measured his workstation precisely: a 38‑inch slightly curved monitor placed about 23 inches from his eyes at center and roughly 29 inches to the edges.

Before the exam he explicitly told clinic staff that he looks up to his monitor rather than down as when reading, provided the exact distances, and requested two single‑vision pairs — one for distance and one tuned for computer work-specifying that he did not want progressives or bifocals. Despite that, the clinic recorded lens powers that placed near or reading correction in the lower portion of the issued computer pair rather than optimizing for an intermediate working distance around his monitor.

Gewirtz describes the standard refraction process as a rapid sequence of subjective lens swaps that often occurs under time pressure; this method can produce reliable distance prescriptions but miss intermediate‑distance needs. After months of increasingly strained mid‑distance vision with the computer pair, he fed the written prescription numbers into ChatGPT, Claude and Gemini to compare the lens powers against his measured workstation distances. All three models agreed on the diagnosis: the numbers corresponded to reading‑oriented correction, not single‑vision intermediate lenses for monitor use.