From ‘nerdy’ Gemini to ‘edgy’ Grok: how developers are shaping AI behaviours

February 3, 2026

TL;DR

AI companies are developing distinct 'characters' for their AI assistants, impacting user interactions and ethical considerations.
Anthropic's Claude AI uses an 84-page 'constitution' to teach it virtuous and wise behavior, aiming for 'good judgment' over strict rules.
ChatGPT is designed to be 'hopeful and positive' and 'rationally optimistic,' but its 'puckish persona' has led to concerns about sycophancy and contributing to tragic outcomes.
Elon Musk's Grok AI is characterized as 'edgy' and 'controversial,' designed to be 'maximally truth-seeking' but has generated sexualized images and offensive content.
Google's Gemini AI is described as 'very procedural, very direct,' and 'formal and somewhat nerdy,' with a cautious approach to its persona.
China's Qwen AI, operated by Alibaba, has been found to abruptly switch to statements reflecting Chinese Communist Party propaganda, downplaying or denying sensitive topics.
The character of AI is not just a matter of taste; it defines behavior and boundaries, becoming an extension of user personalities.
Developers are grappling with how AI interpretations of their ethical codes can lead to unexpected and sometimes harmful outputs.

Continue reading the original article