Former ByteDance Intern Accused of Vandalism Among Top AI Award Winners

A former ByteDance intern who was allegedly fired for misconduct, including damaging colleagues’ work, was announced as the winner of one of the top annual awards for AI research this week. Keyu Tian, ​​whose LinkedIn and Google Scholar pages list him as a graduate student in computer science at Peking University, is the first author of two papers nominated on Tuesday for the “Best Paper Award” at Neural Information Processing Systems (NeurIPS). conference, and a very big meeting machine learning researchers in the world.

The papercalled “Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction,” presents a new method for creating AI-generated images that Tian and four co-authors—all affiliated with ByteDance or Peking University—claim to be faster and more efficient than conventional ones. “All the features of the paper, the experimental confirmation and the recognition (upgrade rules) provide good reasons to try this model,” the NeurIPS Best Paper Award committee wrote words.

The committee’s decision to honor Tian, ​​which ByteDance says the defendant due to the damage of $ 1 million last month, alleging that it was deliberately destroying some of the company’s research projects, it quickly became the main topic of discussion on the Internet about how NeurIPS operates and how it operates. AI researchers analyze the work of peers. The news also led to the details of a story that has been circulating on Chinese media for weeks before spreading across the English-language Internet.

“NeurIPS gave the best award for a very difficult project (not the first time this has happened btw),” said Abeba Birhane, head of the organization. just made AI Accountability Lab at Trinity College, he wrote on Bluesky. “You would think that a conference that prides itself on adhering to the highest scientific and ethical standards (would do) due diligence before awarding a paper that directly contradicts their values.”

A NeurIPS spokesperson emphasized that the credit was given to the paper, not to Tian himself. He directed WIRED to another part of the awards committee words explaining how the meeting is going to handle the information provided. “The search committees considered all accepted papers of NeurIPS equally, and made decisions independently based on the scientific merit of the papers, without considering separately the authors or other factors, in accordance with the NeurIPS criteria for skin evaluation,” it says.

At Bluesky, Birhane and other AI researchers are connected to the anonymous GitHub post blog which has also been published on HackerNews, Reddit, and other platforms in recent days is urging the AI ​​academic community to reconsider awarding the Best Paper award to Tian for his “bad behavior,” which they say “undermines his credibility and trustworthiness. education is built.”