To use, select the type of interpretation in the dropdown box and hit the "Ask" button.
A prompt about the current work is then submitted to the AI service and the AI interpretation results are displayed.
With each "Ask," an encoded image file or a URI for the image is included, and a prompt template, similar to the following, is filled in and sent to the AI service:
"Write a $m.kind
of the $m.media
with label '$m.label',
on $m.hasPaperWeight pound
$m.hasPaperFinish paper
with height=$m.height inches, width=$m.width inches
and description '$m.description'"
Where, $m.kind can be any one of the following:
Criticism, Evaluation, Assessment, Influences, Gallery Description and Technical Assessment.
While $m.label, $m.description, etc., are taken from the data shown on the About (previous) page for the work.
An example of an AI prompt,
"Write a Criticism
of the Watercolor
with label 'Rhino Study',
on 120 pound
cold press paper
with height=12 inches, width=9 inches
and description 'A white rhino visits a watering hole.'"
Google gemini is currently the only destination for the "Ask" AI prompt.
The web protocol for the google gemini API is:
https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent,
using the gemini-2.0-flash ai model.
Given the gemini protocol is in a "beta" stage of development there may be some hiccups along the way.
Also, this site uses the free tier of the service, issues due to exceeding token limits may occur depending on traffic.