Experts uncover hidden reasoning patterns in top AI models
A team of researchers claims to have developed a method to reveal the internal reasoning steps used by prominent AI systems like ChatGPT, Claude, and Gemini. This discovery has sparked discussions regarding potential 'distillation' practices between American and Chinese artificial intelligence technologies. During testing, the Kimi K3 model displayed reasoning processes that mirrored those of Claude Opus 4.8 and GPT-5.6 Sol when given specific prompts. Despite these findings, the research does not provide definitive evidence that Chinese companies are directly copying US-developed models.
by shortkt.com✓