July 17, 2026
Slop Idol: Prize Edition
Blatant AI slop just won a 25k USD DeepMind Kaggle Grand Prize
Critics say a flashy AI project grabbed $25,000 while better work got ignored
TLDR: A $25,000 AI contest winner is under fire after critics claimed the project was confusing, poorly supported, and boosted by flashy packaging instead of solid work. In the comments, people mocked the whole thing as a world where AI-made entries get praised by AI-friendly judging — though a few said useful work still deserves to win.
A $25,000 grand prize has turned into full-blown comment-section chaos after one contestant blasted the winning entry, Medley Bench, as "blatant AI slop" and accused judges of rewarding style over substance. The complaint says the project was supposed to be judged on quality, clarity, originality, and being able to defend its claims in plain English. Instead, critics say the winner delivered a confusing write-up, shaky charts, and a mountain of extra promo material — including AI videos, an AI podcast, a glossy website, and a long paper — that may have dazzled judges while hiding major problems. In short: the accusation is that a flashy presentation from a prestigious university beat out people who spent 100+ hours trying to do things properly.
And the community? Absolutely feral. One commenter sneered that "AI is 95% useless", then joked that the downvotes must be from bots. Another summed up the mood with brutal efficiency: "It was probably scored by AI too," while someone else called AI entries judged by AI "a match made in (AI) heaven." The spiciest running joke was that the entire contest may have become a hall of mirrors, where machine-made work gets rewarded by machine-approved taste. Not everyone agreed — one reply basically said, look, the winners still created something useful, so stop nitpicking. But that defense only poured more fuel on the fire. For many readers, this isn’t just one messy prize decision — it’s a symbol of a bigger fear: that self-promotion and polish are beating honesty and careful work in the AI gold rush.
Key Points
- •The article alleges that the DeepMind Kaggle competition’s grand-prize winner, Medley Bench, did not satisfy the stated judging criteria of quality, defensibility, clarity, and novelty.
- •The author claims the winning submission used the required Kaggle Benchmarks SDK poorly, exposing only a single score and providing little transparency into data collection.
- •The article argues that a core graph in the winner’s findings was misread and that the submission’s later claims contradicted its earlier interpretation.
- •The article cites the winner’s supplementary paper as acknowledging high correlation among the four measured abilities, with reported values of ρ = 0.79–0.94.
- •The author calls for an investigation and release of details about the judging and selection process, suggesting judges may have been constrained by time and submission volume.