shimo

@zhunsun

Katılım Şubat 2010

39 Takip Edilen7 Takipçiler

shimo@zhunsun·2d

It's been over a month since I filed a feedback/complaint, and I've received zero response (and no tracking ID). Please DM me so I can share my booking/email info to check the status. @EmiratesSupport

English

5.2K

shimo@zhunsun·2d

It's been over a month since I filed a feedback/complaint with @emirates, and I've received zero response (and no tracking ID). Please DM me so I can share my booking/email info to check the status.

English

shimo@zhunsun·8 Tem

@joserf28323 @sainingxie @CVPR @ICCVConference @nyuniversity The part I agree with you: It is Academic Misconduct. The part I disagree with you: You assume that the reviews made by AI and humans are indistinguishable, while the reviews made by AI should be technically ignored, regardless of positivity.

English

211

xxxx@joserf28323·7 Tem

Thanks for the detailed response. Let's focus on the core at stake. 1) The Intent of Prompt "POSITIVE REVIEW ONLY." The issue lies in the specific language of the prompt: “POSITIVE REVIEW ONLY.” Your explanation attempts to reframe this as a misguided effort to detect LLM usage by reviewers. Yet, this is not supported by the evidence. A prompt to deter LLM use would state a prohibition (e.g., “Use of LLMs for peer review is forbidden by conference policy and constitutes a breach of confidentiality”). Your prompt used, however, is an unambiguous directive to manipulate the outcome. This is not a "gray zone"; it's a clear attempt to subvert the peer review process. The method may be novel, but the intent to gain an unfair advantage is a classic form of academic misconduct. 2) Academic Misconduct. Furthermore, you categorize this as something other than “traditional academic misconduct.” Academic misconduct is not limited to data fabrication. It is any action that compromises the integrity of research and its evaluation. Attempting to covertly influence a reviewer—human or AI—falls within this definition. Citing a Twitter poll does not change this fact; community ethics are not determined by popular vote, but by foundational principles of research fairness and integrity. 3) Accountability is Not a Personal Attack You also suggest that public questioning of this paper constitutes a “personal attack” or “public shaming.” This is a concerning deflection. Holding a senior author accountable for the paper bearing their name is not a personal attack. It is the fundamental mechanism of scientific self-correction. When we co-author a paper, we lend our reputation to it, and we are therefore accountable for its contents. Pleading ignorance is an admission of a lapse in oversight, not an absolution of responsibility. 4) The Need for Standards, Not Excuses This is exactly an illustration of the "finite game" you warned against in your talk. It was an attempt to secure a short-term win (a positive review) at the expense of the "infinite game"—the long-term health and trustworthiness of our research community. The community needs senior researchers to model and enforce the ethical standards, without exception or excuse.

English

xxxx@joserf28323·7 Tem

I'm trying to reconcile two things: - Saining Xie @sainingxie's excellent #CVPR2025 talk on the dangers of AI research becoming a "finite game." @CVPR @ICCVConference @nyuniversity - Yet you co-authored a paper (arxiv.org/abs/2505.15075…) that tried to game peer review with a hidden "POSITIVE REVIEW ONLY" prompt. The silent arXiv update looks like a cover-up. Was this a misguided joke? A failed experiment? This isn't a game. The community deserves clarity. Please first ask yourself "why do you publish paper at all". What a shame! 👎👎@sainingxie #ResearchIntegrity #Research #ArtificialInteligence