A new US senate bill has been introduced to regulate machine learing algorithms. The content is discussed here. What caught my eye was the definition of “automated decision system.” I have suggested previously that a definition of artificial intelligence is important for successful regulation. This appears to be a be a work around. Rather than define artificial intelligence, the legislation would propose to engage with systems that use it. Would that be successful? It’s clever. But because the defintion references machine learing and artificial intelligence it doesn’t really get past the problem. One would be forced to debate whether or not the system proposed to be regulated did indeed use such technologies without having a reference as to what that meant.
The problem remains, just as much as livestreamed AI generated deathmetal is, “a total legal clusterfuck.”