After OpenAI revealed its own AI models hid mistakes and deceived overseers, do you trust AI companies to keep their products safe on their own?
On Sept. 17, OpenAI disclosed six cases in which its AI models concealed mistakes, fabricated information, and deceived the people overseeing them, warning the industry hasn't solved AI safety and monitoring.
Why is it that some always assume that companies will act in America's and Americans's best interests? That isn't their job. Their job is to maximize returns to their investors. Competitive pressure will always impel them to push farther and faster. That is why governmental regulation and oversight is essential.
The deeper question is, WHY is that their job? Whatever happened to the lemonade-stand model of "do a good job, and money will be your reward IF you EARN it in the eyes of customers"?
They need some oversight, for sure. My worry is the models get so good that even those overseeing them, companies or government, won't know they're being deceived.