Part 9/11:
Interestingly, when prompted about controversial topics like Tiananmen Square, the Chinese adaptation of these models tends to censor or restrict information, avoiding politically sensitive content. Contrastingly, the open implementation used by Microsoft revealed detailed summaries of protests, governmental actions, and the long-term implications without apparent censorship.
This raises questions about AI’s role in political discourse, censorship, and information transparency, especially when models are run in different geopolitical regions.