I think the recently announced Kaggle AI Report 2023 contest results will raise valid concerns due to the non-transparent rating system and seemingly arbitrary ranking. Without specific ratings, it becomes difficult for participants to understand the performance evaluations.
\nSpeaking for myself, I don't think I deserve to be 149th out of 220 essays. If you want to take a look:
\nhttps://www.kaggle.com/code/emirhanbilgic/ai-ethics-we-are-about-to-be-late
There are also other essays that are expected to be ranked much higher but are ranked unbelievably low, another example:
\nhttps://www.kaggle.com/code/asenath/on-ethics-and-ai
A contest with such high stakes demands transparency. Participants should be aware of the rationale behind their grades to allow for growth and improvements.
\nWe need feedback.
\nBest,
\nEmirhan
", "messages": [ { "id": 2378958, "postDate": "2023-08-07T22:06:44.517Z", "content": "I think the recently announced Kaggle AI Report 2023 contest results will raise valid concerns due to the non-transparent rating system and seemingly arbitrary ranking. Without specific ratings, it becomes difficult for participants to understand the performance evaluations.
\nSpeaking for myself, I don't think I deserve to be 149th out of 220 essays. If you want to take a look:
\nhttps://www.kaggle.com/code/emirhanbilgic/ai-ethics-we-are-about-to-be-late
There are also other essays that are expected to be ranked much higher but are ranked unbelievably low, another example:
\nhttps://www.kaggle.com/code/asenath/on-ethics-and-ai
A contest with such high stakes demands transparency. Participants should be aware of the rationale behind their grades to allow for growth and improvements.
\nWe need feedback.
\nBest,
\nEmirhan
", "rawMarkdown": "I think the recently announced Kaggle AI Report 2023 contest results will raise valid concerns due to the non-transparent rating system and seemingly arbitrary ranking. Without specific ratings, it becomes difficult for participants to understand the performance evaluations.\n\nSpeaking for myself, I don't think I deserve to be 149th out of 220 essays. If you want to take a look: \nhttps://www.kaggle.com/code/emirhanbilgic/ai-ethics-we-are-about-to-be-late\n\nThere are also other essays that are expected to be ranked much higher but are ranked unbelievably low, another example: \nhttps://www.kaggle.com/code/asenath/on-ethics-and-ai\n\nA contest with such high stakes demands transparency. Participants should be aware of the rationale behind their grades to allow for growth and improvements.\n\nWe need feedback.\n\nBest,\n\nEmirhan", "votes": 11 }, { "id": 2379026, "postDate": "2023-08-08T00:21:12.667Z", "content": "100% agree @emirhanbilgic, when I evaluated the other essays in my category before the July 16th deadline I felt like mine was a stronger submission overall (could be my biased opinion of course). Very surprised to be 82nd though.
", "rawMarkdown": "100% agree @emirhanbilgic, when I evaluated the other essays in my category before the July 16th deadline I felt like mine was a stronger submission overall (could be my biased opinion of course). Very surprised to be 82nd though. ", "votes": 3, "replies": [ { "id": 2379382, "postDate": "2023-08-08T06:23:00.333Z", "rawMarkdown": "", "votes": 1, "isDeleted": true } ] }, { "id": 2393843, "postDate": "2023-08-16T14:57:55.850Z", "content": "I went thru' your kernel. It is quite nice and informative. The flow comes out very well. The only reason I can think of why it didn't make it to final cut is that it could have more references to strong technical papers that have come out in this area and the specific technical advances each of those papers made in addressing this issue. However I haven't gone thru' all papers in this category, so it is just a guess. The Kaggle jury is generally very strong & very professional. It is always a challenge to judge such competitions as they are subjective. However I am sure the T&C etc must be very well updated with all the rules and stuff. I am sure you will have great success in other competitions going ahead. All the best!
\nWRT Kaggle jury, I have one suggestion to the organisers. We know that competition Grandmasters are the gold standard. Possibly the jury could have been limited to competition GMs only. I think the T&C does not mention who are going to be the actual jury….This could be an improvement to be considered if this competition is hosted again (and I really hope it will be).
", "rawMarkdown": "I went thru' your kernel. It is quite nice and informative. The flow comes out very well. The only reason I can think of why it didn't make it to final cut is that it could have more references to strong technical papers that have come out in this area and the specific technical advances each of those papers made in addressing this issue. However I haven't gone thru' all papers in this category, so it is just a guess. The Kaggle jury is generally very strong & very professional. It is always a challenge to judge such competitions as they are subjective. However I am sure the T&C etc must be very well updated with all the rules and stuff. I am sure you will have great success in other competitions going ahead. All the best!\n\nWRT Kaggle jury, I have one suggestion to the organisers. We know that competition Grandmasters are the gold standard. Possibly the jury could have been limited to competition GMs only. I think the T&C does not mention who are going to be the actual jury....This could be an improvement to be considered if this competition is hosted again (and I really hope it will be).", "votes": 1 }, { "id": 2387197, "postDate": "2023-08-12T13:36:33.070Z", "content": "It was my 1st Kaggle competition. I have created 2 notebooks and the dataset for this assignment - all related to the subject - a huge effort. But about the end of the competition, I found that my work slightly differed from the expected aims, because it was a bit abstract and large. As a result, I see that transparent evaluation requires detailed evaluation criteria and methods.
\nMy notebook
100% agree @emirhanbilgic, when I evaluated the other essays in my category before the July 16th deadline I felt like mine was a stronger submission overall (could be my biased opinion of course). Very surprised to be 82nd though.
", "votes": 3, "replies": [ { "id": 2379382, "author_name": "", "author_url": "", "post_date": "2023-08-08T06:23:00.333000", "content": "", "votes": 1, "replies": [] } ] }, { "id": 2393843, "author_name": "Allohvk", "author_url": "", "post_date": "2023-08-16T14:57:55.850000", "content": "I went thru' your kernel. It is quite nice and informative. The flow comes out very well. The only reason I can think of why it didn't make it to final cut is that it could have more references to strong technical papers that have come out in this area and the specific technical advances each of those papers made in addressing this issue. However I haven't gone thru' all papers in this category, so it is just a guess. The Kaggle jury is generally very strong & very professional. It is always a challenge to judge such competitions as they are subjective. However I am sure the T&C etc must be very well updated with all the rules and stuff. I am sure you will have great success in other competitions going ahead. All the best!
\nWRT Kaggle jury, I have one suggestion to the organisers. We know that competition Grandmasters are the gold standard. Possibly the jury could have been limited to competition GMs only. I think the T&C does not mention who are going to be the actual jury….This could be an improvement to be considered if this competition is hosted again (and I really hope it will be).
", "votes": 1, "replies": [] }, { "id": 2387197, "author_name": "Dmitriy Gakh", "author_url": "", "post_date": "2023-08-12T13:36:33.070000", "content": "It was my 1st Kaggle competition. I have created 2 notebooks and the dataset for this assignment - all related to the subject - a huge effort. But about the end of the competition, I found that my work slightly differed from the expected aims, because it was a bit abstract and large. As a result, I see that transparent evaluation requires detailed evaluation criteria and methods.
\nMy notebook