Experimental browser for the Atmosphere
The fact that the research team is responding to this controversy with additional bullshit is a huge red flag regarding the integrity of their work. If they won't admit they broke rules that they clearly broke, what reason is there to believe a higher standard will be applied to a published paper?
Apr 27, 2025, 5:09 PM
{ "uri": "at://did:plc:7n2gzbzn4xus4nghzfpjgli3/app.bsky.feed.post/3lnsqkbf5422f", "cid": "bafyreifuyo7r4v6r4rozzkveh6kqscpkwte7w46kj6amo2r2jtqtf2nodi", "value": { "text": "The fact that the research team is responding to this controversy with additional bullshit is a huge red flag regarding the integrity of their work.\n\nIf they won't admit they broke rules that they clearly broke, what reason is there to believe a higher standard will be applied to a published paper?", "$type": "app.bsky.feed.post", "embed": { "$type": "app.bsky.embed.recordWithMedia", "media": { "$type": "app.bsky.embed.images", "images": [ { "alt": "CMV’s rules state that “The use of AI text generators (including, but not limited to ChatGPT) to create any portion of a post/comment must be disclosed and substantial human-generated content included; failure to do so is a Rule 5 violation”. Specifically, this rule falls under the subreddit’s broader policies against “low-effort” and “low-quality” responses (Rule 5: “Responses must contribute meaningfully to the conversation”). While we acknowledge that our intervention did not uphold the anti-AI prescription in its literal framing, we carefully designed our experiment to still honor the spirit behind Rule 5. In particular, we developed a comprehensive posting pipeline including multiple rounds of automated review and human oversight to ensure high-quality, contextually relevant AI-generated contributions. We note that our comments were consistently well-received by the community, earning over 20,000 total upvotes and 137 deltas. Only two of our comments were removed under Rule 5, suggesting we met expectations for effort and relevance.", "image": { "$type": "blob", "ref": { "$link": "bafkreibg5o4hn6ydlzlo53mzxfad3u2rimluxzj3s3j2of5kmj2mngrcwu" }, "mimeType": "image/jpeg", "size": 434053 }, "aspectRatio": { "width": 960, "height": 832 } }, { "alt": "ChangeMyView prohibits bots. You violated the community rules!\n\nCMV’s rules state that “Bots, novelty and spam-only accounts are also unilaterally banned”, under a paragraph listing \"blatantly unacceptable behavior\" related to \"Disrupting the Subreddit\". While our posting pipeline involved some degree of automation, our research accounts do not meet the conventional definition of “bots”. Unlike typical bots, which autonomously generate large volumes of content without human oversight, our accounts posted a very modest number of comments, averaging only 10-15 per day (a negligible portion of the subreddit’s activity, which averages about 7,000 comments per day). Importantly, while the text was generated using LLMs, every single comment was reviewed and ultimately posted by a human researcher, providing substantial human oversight to the entire process. Given these considerations, we consider it inaccurate and potentially misleading to consider our accounts as \"bots\", and we believe that these accounts should not fall within the scope of this rule, which was written in a completely different spirit to prevent large operations from disrupting the subreddit. Our goal was never to disrupt, spam, or dilute the quality of CMV’s conversations: in fact, we specifically designed our process to align with the subreddit’s values by ensuring every comment contributed meaningfully to the discussion.", "image": { "$type": "blob", "ref": { "$link": "bafkreigok3yx3jk74qhjtxiyqvt2bzgr4sa7ufuob4ynhzobhukqfhbgyy" }, "mimeType": "image/jpeg", "size": 658097 }, "aspectRatio": { "width": 958, "height": 974 } } ] }, "record": { "$type": "app.bsky.embed.record", "record": { "cid": "bafyreigmlt4hwyehhe7mxbc7d2a6daq3nzdkutgifl25zhtkvmz666bgpa", "uri": "at://did:plc:mkzljt6pfd2tzf5oj7hz5v5w/app.bsky.feed.post/3lnqqhzpvzs2c" } } }, "langs": [ "en" ], "createdAt": "2025-04-27T17:09:13.884Z" } }