A former ByteDance intern who was allegedly dismissed for skilled misconduct, together with sabotaging colleagues’ work, was introduced as a winner of one of the vital prestigious annual awards for AI analysis this week. Keyu Tian, whose LinkedIn and Google Scholar pages checklist him as a grasp’s pupil in laptop science at Peking University, is the primary creator of one among two papers chosen Tuesday for the principle Best Paper Award on the Neural Information Processing Systems (NeurIPS) convention, the most important gathering of machine-learning researchers on the planet.
The paper, titled “Visual Autoregressive Modeling: Scalable Image Generation by way of Next-Scale Prediction,” presents a brand new methodology for creating AI-generated photos that Tian and 4 coauthors—all affiliated with both ByteDance or Peking University—declare is quicker and extra environment friendly than its predecessors. “The total high quality of the paper presentation, experimental validation and insights (scaling legal guidelines) give compelling causes to experiment with this mannequin,” the NeurIPS Best Paper Award committee wrote in an announcement.
The committee’s resolution to grant the honour to Tian, whom ByteDance reportedly sued for over $1 million in damages final month, claiming deliberate sabotage of different firm analysis tasks, rapidly turned the main target of wider discussions on-line about how NeurIPS is run and the way in which high AI researchers consider the work of their colleagues. The information additionally prompted the small print of a scandal that had been brewing on Chinese social media for weeks to lastly spill over onto the English-language web.
“NeurIPS gave greatest paper award to an excellent problematic work (not first time this has occurred btw),” Abeba Birhane, head of the newly shaped AI Accountability Lab at Trinity College, wrote on Bluesky. “You’d suppose a convention that prides itself on upholding the very best scientific & moral normal would [do] due diligence earlier than they provide the award to a paper that instantly contradicts their values.”
A spokesperson for NeurIPS confused that the honour was given to the paper, to not Tian himself. They directed WIRED to a portion of the award committee’s assertion explaining how the convention evaluates paper submissions. “The search committees thought-about all accepted NeurIPS papers equally, and made selections independently based mostly on the scientific benefit of the papers, with out making separate issues on authorship or different components, in line with the NeurIPS blind assessment course of,” it reads.
On Bluesky, Birhane and different AI researchers linked to an nameless GitHub weblog put up that additionally circulated on HackerNews, Reddit, and different platforms in latest days urging the educational AI neighborhood to rethink granting the Best Paper honor to Tian due to his “critical misconduct,” which it says “essentially undermines the core values of integrity and belief upon which our tutorial neighborhood is constructed.”
