Dev fires back at 'benchmaxxed AI' critics: humans gamed exams in school too
gnukeith · x · 2026-10-28
X user gnukeith mocks the common criticism that AI models are 'benchmaxxed': humans in school did exactly the same — optimizing for exams and even cheating — so why single out models for optimizing against benchmarks?
A one-liner that lands a sharp point about the double standard in the 'models just game benchmarks' debate.
More from Fun
- Baidu's open OCR model digitizes all 270 pages of an 1899 Merck manual on one GPU — MaziyarPanahi · 2026-10-29
- Complexity theory joke: OpenAI's 'galactic algorithm' only runs exponential subroutine past n > 10^10^25 — brandondamos · 2026-10-29
- Company Admits Provoking X Community After Botched AI Release, Credibility 'Speedrun to Zero' — TheZachMueller · 2026-10-29
- AI Circle Meme: Adam Grant's 2x2 Matrix for 'Human Connection' Goes Around — floguo · 2026-10-29
- IT vendor lists $60M AI supercomputer with 9,600 A100 GPUs on eBay — free shipping included — giveen · 2026-10-29
- Polymarket amplifies claim of an 'evil neuron' in Qwen that triggers violent outputs — Polymarket · 2026-10-29