1.26M blind votes show users judge AI more harshly each month — models aren't declining
rohanpaul_ai · x · 2026-11-14
An analysis of 1,260,967 blind Chatbot Arena votes (May 2023–Aug 2024) on frozen model pairs finds 'both are bad' votes rise 1.71 percentage points per month off a 17.1% mean (1.91 pts/month in 2024), while stronger models' win shares stay flat. Conclusion: user standards are rising, not model quality declining — with implications for leaderboards and human feedback used in training. Paper: 'AI Isn't Getting Dumber, We're Getting Pickier.'
Related event: 1.26M Blind Votes Show Humans Are Getting Tougher on LLMs(2 posts)→
More from Models
- Developer downgrades Claude Ultra to Pro and moves all work to Codex over 5-hour rate limits — umeshonai · 2026-11-14
- Pangram details per-domain false positive rates (1 in 10,000), enabling profitable AI detection — soldni · 2026-11-14
- Blogger predicts GPT-6.2 launches before Nov 19, floats three reasons for the delay — kimmonismus · 2026-11-14
- Dev observes Fable 5.2 is making design decisions closer to his own than Opus 5.5 — adonis_singh · 2026-11-14
- 5,400 real-robot rollouts benchmark Claude Fable 5.2 at 65%, ahead of Gemini and GPT — ricklamers · 2026-11-14
- Google Antigravity adds Gemini 4 Hydrogen and Orchestrator for long-horizon agent tasks — minchoi · 2026-11-14