homogeneity

April 24, 2011 - Earlier this month OpenAI released the GPT-4.1 artificial intelligence model and claimed that it performed well in following instructions. However, the results of several independent tests have shown that GPT-4.1's alignment (i.e., reliability) appears to have declined when compared to OpenAI's previously released models. According to 1AI, normally, when OpenAI releases a new model, it releases a detailed technical report containing the results of both first- and third-party security evaluations. But this time, for GPT-4.1, the company didn't...