People need a bad capitalist, AI creates a food delivery rumor Last week there was something quite magical. The CEOs of two food delivery giants in the United States, one with a net worth of 2.7 billion US dollars and the other in charge of the world's largest ride hailing platform, stayed up early Saturday morning and wrote a short essay online to prove their innocence. The cause was an anonymous post on Reddit. The person who posted claimed to be a backend engineer for a large food delivery platform, got drunk, and went to the library to expose it using public WiFi. The content is roughly as follows: The company will analyze the situation of ride hailing drivers and give them a "despair score", with drivers who lack money being less likely to receive good orders; The so-called priority delivery for takeout is false, as regular orders will be delayed; All kinds of 'driver welfare fees' were not given to drivers at all, and were used to lobby Congress to deal with unions .. The ending of the post is very realistic: I was drunk and angry, so I want to expose it. To portray oneself as a whistleblower who "big companies use algorithms to squeeze drivers". The post was posted for three days, received 87000 likes, and went viral on the Reddit homepage. Some people also took screenshots and sent them to X, gaining 36 million exposures. You should know that there are only a few big players in the US food delivery market. The post did not name them, but everyone is guessing who they are. DoorDash CEO Tony Xu couldn't sit still and tweeted that we didn't do this. Whoever dares to do this will be fired. Uber's COO also stepped forward to respond, 'Don't believe everything you see online.'. DoorDash even posted a five point statement on its official website, refuting the contents of the leak one by one. These two companies, with a combined market value of over 80 billion US dollars, were cleared up overnight by an anonymous post. Then, it was surprisingly proven that this post was created by AI. It was exposed by Casey Newton, a journalist from overseas technology media platform Platformer. He contacted the whistleblower of this post, who immediately sent an 18 page "internal technical document" with a scholarly title called "AllocNet-T: High Dimensional Temporary Supply State Modeling". Translated roughly as' high-dimensional temporal supply state modeling '. Each page is marked with a "confidential" watermark, signed by Uber's "Market Dynamics Group · Behavioral Economics Department". The content is about explaining how the model in the Reddit post that gives drivers a "despair score" is calculated. There are architecture diagrams, mathematical formulas, and data flow charts inside .. (Screenshot of a fake paper, at first glance, it looks just like the real thing) Newton said that this document initially deceived him. Who would spend effort forging 18 pages of technical documents to lure a journalist? But now it's different. This 18 page document can be generated in just a few minutes using AI. At the same time, the whistleblower also handed over a photo of their Uber employee ID code to the reporter, indicating that they were indeed working inside. Out of curiosity, journalist Newton threw the employee ID into Google Gemini for verification, and Gemini claimed that the image was generated by AI. It can be recognized because Google has embedded an invisible watermark called SynthID in the content produced by its own AI, which is invisible to the naked eye but can be detected by machines. Even more outrageous is that the employee ID bears the logo of "Uber Eats". An Uber spokesperson confirmed that we do not have employee IDs for the Uber Eats brand, and all badges only print Uber. Obviously, this false whistleblower has not even figured out who to blacklist. When reporters request social media account information such as LinkedIn for further verification, The whistleblower directly deleted their account and ran away. Actually, what we want to talk about is not that AI can fake, it's not new. What we want to talk about more is: why are tens of millions of people willing to believe an anonymous disclosure post? In 2020, DoorDash was sued for using tips to offset the driver's base salary, resulting in a compensation of $16.75 million. Uber once developed a tool called Greyball specifically designed to evade regulation. These are true stories. You can easily find a subconscious agreement that the platform is not a good thing, and this judgment is accurate. So when someone says' food delivery platform algorithms exploit drivers', their first reaction is not 'is this true', but 'it's true'. Fake news can run because it looks like what people already believe in their hearts. What AI does is to reduce the cost of this' like 'to almost zero. There is another detail to this story. Detecting scams relies on Google's watermark detection. Google makes AI, and Google also makes AI detection tools. But SynthID can only check Google's own AI. We were able to catch it this time because the counterfeiter happened to use Gemini. If it were another model, you might not be so lucky. So this case is not so much a technical victory as it is: The other party made a low-level mistake. Previously, a Reuters survey showed that 59% of people were worried about not being able to distinguish between truth and falsehood online. The clarification tweet from the CEO of a food delivery company has been viewed by hundreds of thousands of people, but how many of them firmly believe that it is public relations or a lie? Although the fake disclosure post has been deleted, there are still people cursing the food delivery platform in the comment section. The lie has run half the world, but the truth is still tying shoelaces. Think about it again, what if this post is not about Uber, but about Meituan or Ele.me? What are the 'despair scores', what are the' algorithms used to squeeze riders', what are the 'welfare fees not given to riders'. When you see these accusations, is your first reaction emotional identification? Do you remember the article 'Delivery Riders, Trapped in the System'. So the question is not whether AI can falsify. The question is, when a lie looks like something that everyone has long believed in, is truth still important? The person who deleted the account and ran away, what's the plan? I don't know. I only know that he found an emotional outlet and poured a bucket of AI generated fuel into it. It's on fire. As for whether the firewood is real or fake, who cares? In fairy tales, Pinocchio's nose grows longer when he lies. AI doesn't have a nose. Article link: https://www. (techflowpost.com)/zh-CN/article/29848