“AI outputs can be misleading or wrong”

by Steven Garrity

AI features keep popping up in all of the digital products I use. Figma, Photoshop, Notion, Zoom. These products tend to prompt acceptance of vague privacy terms that may or may not feed your private/company data into the AI slurry.

Figma added some AI generation tools into their FigJam product under a “Beta” caveat. I’m both impressed at how candid they are and troubled at how they are comfortable shipping a feature that requires such a warning:

“AI outputs can be misleading or wrong”

— FigJam product, which uses AI…
Screenshot of software interface with text input to generate material with AI and a warning that reads 'AI outputs can be misleading or wrong'

Also, if you ask the Photoshop’s AI-powered “Generative Fill” feature to draw something “with a transparent background”, it renders a messy checker-board background that is sometimes used to illustrate transparency.

Two images of trees with an uneven checker-board pattern in the background.

Lots of power, lots of potential, lots of problems.

 

Leave a Reply

Your email address will not be published. Required fields are marked *

To respond on your own website, enter the URL of your response which should contain a link to this post's permalink URL. Your response will then appear (possibly after moderation) on this page. Want to update or remove your response? Update or delete your post and re-enter your post's URL again. (Find out more about Webmentions.)