#trending: Redditor alleges GrabFood app using AI-generated food images, netizens say it's 'deceiving'
Food menu pictures for a Japanese restaurant on the GrabFood application that online users believe are generated by artificial intelligence software.
SINGAPORE — Online users have decried the alleged use of food images generated by artificial intelligence (AI) on food delivery platform GrabFood, after someone posted photos on online forum Reddit's discussion thread last Saturday (May 6).
While browsing some restaurants on GrabFood, Reddit user “Marlix6754” noticed that one Japanese shop from which the user had ordered before had changed its menu images.
Screenshots from the GrabFood mobile application show a list of various Japanese dishes such as chirashi don, teriyaki chicken don and fried chicken gyoza. However, upon closer inspection of the photos, one may notice some fine print that read: “Image generated by Grab. Actual product might differ.”
This discovery led the Reddit user to question if GrabFood was trialling the use of AI-generated pictures for some of its merchants.
“This restaurant (did) not use their own images in the past, so my guess is this is AI. The images look very unappetising to me though,” the user wrote.
TODAY checked the GrabFood app and found the text, "Image generated by Grab. Actual product might differ", on several food-and-beverage items from various restaurants and eateries.
Many Redditors in the discussion thread agreed that the pictures looked like they had been created by AI and were turned off by the results, commenting that the food looked “weird”, “sterile” and even “nauseating”.
One user also pointed out discrepancies in the size of the ingredients in the images. For instance, the ikura or salmon roe pictured on the salmon ikura don are as big as the yolk of a salted egg.
The AI image generator supposedly used by GrabFood also seemed to have taken the term “truffle fries” literally, producing a picture with a whole truffle sitting on a bed of shoestring fries.
One user quipped: “None of these pictures show anything I want near my mouth.”
Some Redditors thought that the use of AI-generated images instead of pictures that show the actual products could be misleading to customers.
One person wrote: “Rather deceiving though, as not everybody will read the fine print… I (would) rather (have) no image than an image indicating a chunky salmon, but later (getting) a skinny slice.”
A few users added that the food delivery platform could improve on the quality of the images with the use of better AI software or more detailed prompts.
In response to TODAY's queries, Grab confirmed that its food delivery platform is conducting a trial to study if AI-generated photos could be a feasible solution for merchant-partners who do not have the resources to take photos for their menus.
"We have received feedback from users that they prefer to see photos in the menu, even if it’s illustrative, so that they can scroll through the menu and select their food items more quickly, compared to reading text only."
The pilot involves a "small pool" of merchant-partners who may choose to opt out of the trial. The merchant-partners may also give feedback if they preferred to replace or remove specific photos.
"We will keep improving and updating our pilot. As part of the process, we are reviewing the photos on an ongoing basis, tracking user feedback, and removing or updating images that can be removed," Grab added.
