Kaedim, a start-up specializing in AI and 3D modeling, is under fire from a serious accusation. It claimed to work with stand-alone software, but instead, they were flesh-and-blood employees working hard in the shadows.
This small Californian start-up offered its customers the conversion of 2D images to 3D models, supposedly through the use of a state-of-the-art AI model. However, it turns out that this “magical” intelligence was not one, but that real artists were in charge of the orders, in unworthy conditions. A hard blow for the company’s image, and one more point to fuel the discourse of the detractors of artificial intelligence.
A gross deception
This story is reminiscent of that of the Mechanical Turk, a hoax organized in the 18the century by the Hungarian writer and inventor Wolfgang von Kempelen. He had claimed to design an automaton capable of playing chess alone against a human player to impress the Empress of Austria Maria Theresa in 1770. Instead, his automaton was actually a mannequin into which a human to be able to play chess games. Thanks to this subterfuge, the so-called machine played and won games for 84 years against illustrious chess players and historical figures such as Bonaparte or Catherine of Russia.
The story of the Kaedim start-up is a little less romantic and will not have lasted as long. Selling its miracle solution mainly to artists in the video game industry, it touted its software as a way to save time for its customers. This promise seemed too good to be true. And indeed, she was.
The deception was exposed by 404 Media, who discovered the pot of roses. Behind this polished facade with the high-tech sauce made in California were actually hiding real artists. The latter were responsible for carrying out the image conversions behind the scenes, working in absolute secrecy for paltry sums. A much less sympathetic picture than that painted by the founder of the start-up, Konstantina Psoma.
Kaedim’s response
First of all, the company denied the facts to 404 Media, but rushed to update its website just after the article appeared. The CEO de Kaedim then backtracked. Until then, she admitted that these people were only there for a simple process of verifying the images generated. She then confessed that her software relies heavily on human labor, changing the blurb on her company’s site. A more than suspicious backpedaling.
Subsequently, Kaedim showed its credentials by committing to offer a fully autonomous version of its service by 2024. It is difficult today to imagine that the company will once again win the trust of customers and investors. While Microsoft eliminated its AI ethics team in March, we are not close to seeing the end of this debate. Like a new El Dorado, we have to believe that some actors are ready for any maneuver, however immoral it may be, to find a place for themselves in the sun.
Sources: Digital Factory, 404 Media
2