> I don't see the problem as long as the actual lawyer can intervene when necessary. There's a big problem. Not all wrongness can be identified in the moment.

> I don't see the problem as long as the actual lawyer can intervene when necessary. There's a big problem. Not all wrongness can be identified in the moment. AI can produce convincing wrongness unintentionally and easily.