I think this is a significant concern: "as researchers and experts pore over its accompanying materials, many have expressed disappointment at one particular feature: that despite the name of its parent company, GPT-4 is not an open AI model." As the article states, it means there's no real way to know whether is honest, reliable or safe. We can't even know whether the contents of the training set were legally obtained, or whether they were gathered out of secret CIA files on all of us. I'm still enthusiastic about the possibilities of AI, but increasingly sceptical of the business models being employed to produce them. "OpenAI was founded as a nonprofit but later became a 'capped profit' in order to secure billions in investment, primarily from Microsoft, with whom it now has exclusive business licenses." OpenAI's Ilya Sutskever might argue that "it just does not make sense to open-source" but it seems to me this refers to 'business sense' rather than 'safety sense' and is much more about extracting value from the community that made it possible rather than giving back to it.
Today: 8 Total: 107 [Direct link] [Share]
