DeepMind/Kaggle $25k Prize Award Disputed
TheWerkmeister · reddit · 2026-07-18
The author disputes the judging results of the Google DeepMind sponsored Kaggle competition “Measuring Progress Toward AGI - Cognitive Abilities”, arguing that a highly flawed submission won the $25,000 prize.
The post focuses on compiling evidence regarding the fairness of the award:
- The goal was to design a new AI benchmark based on cognitive science
- The author claims the analyzed work intended to show an LLM different perspectives on 5 statements and observe if it changed its attitude
- However, the final submission was a massive "collage" far exceeding the required format, filled with inadequate arguments
- The author further questions whether the judges actually read the methods, code, and data carefully
The organizers stated that the review followed the rules and is ultimately a matter of subjective judgment.
More from Safety
- Prompt turns Claude Code into an iPhone backup forensics tool exposing 30-day 'deleted' photo copies — alex_verem · 2026-10-27
- Anthropic reveals it bans customers worth 5-7% of revenue every 9 months — Polymarket · 2026-10-27
- AVERI independently audits Amazon's internal AI safety evals in landmark pilot — Miles_Brundage · 2026-10-27
- US chases AGI dominance while China, EU and ASEAN treat AI as public infrastructure — BubblyExtent1693 · 2026-10-27
- Agent Field Kit launches $29 security runbooks for production AI agents — alvelda · 2026-10-27
- DeepSeek Safety Eval Finds "Over-Attribution": Model Mistakes Ordinary Tasks for Tests — Much_Boysenberry9124 · 2026-10-27