Background

This question is exploring whether AI labs in China would be able to compete with US on building AGI or whether sanctions will prevent them from competing effectively.

Resolution Criteria

This question will be resolved as “Yes” if at least one condition is met before 2030.

Condition 1:

The Chinese AI model passes the Kurzweil version of the Turing test

Condition 2:

The Chinese AI model receives at least 90% score on FrontierMath benchmark and at least 90% score on Humanity’s Last Exam.

Fine Print:

It does not matter to this question if the Chinese AI lab steals model weights or documents from US labs. All that counts for positively resolving this question is if the Chinese AI lab provides access to their AI model to companies and customers.