Technology AI, Science and New Things
The latest and most powerful AI models fundamentally lack transparency, a new study by Stanford University reveals. The research, examining and scoring 10 major AI models for transparency, shows that big tech companies like Google and OpenAI are not divulging crucial information about their models, such as training datasets or risk identification. No company scored above 60 per cent and the study stressed the need for improvement across the board. Most of today’s groundbreaking AI development is occurring behind the closed doors of these companies rather than via public research – OpenAI has not revealed the model size or architecture of their flagship GPT-4 model.
