User Tests Claude: Model Drastically Underestimates 3-4 Year Olds' Cognitive Abilities
TheCharalampos · reddit · 2026-07-30
A user asked Claude to review a treasure hunt plan for a 3-year-old's birthday party and was shocked to find that Claude assumes children of that age are completely incapable. The model incorrectly suggested they cannot understand verbal commands, would choke on random debris, and pummel each other.
After testing across multiple chats, the user confirmed the model consistently underestimates toddlers' capabilities, noting it feels divorced from reality and questioning if this stems from skewed training data or over-alignment.
More from Models
- Trying Grok bot surfaces an unpaid Cursor balance, blocking account merging — haydendevs · 2026-10-16
- TensorFold 1.0.12 runs 120B vision model on 1-2 DGX Sparks, adds Mac image input — HankYeomans · 2026-10-16
- Leaked screenshot shows unannounced Gemini 4 Argon gated to Ultra and paid Pro subscribers — il_-bro · 2026-10-16
- Google: Gemini 4 Argon lands in Antigravity, code submissions up 35%, rollbacks down ~33% — tekbog · 2026-10-16
- Google's "ungrounded beliefs" classifier injections derail Opus 4.1's reasoning chain — liminal_bardo · 2026-10-16
- A week of ChatGPT Pro quota burned in 36 hours as GPT-6 fails to factor a 512-bit RSA key — matthew_d_green · 2026-10-16