Many AI systems these days are adequate counterfeit humans in ways that adding machines and spreadsheets (and advanced systems that reduce to these things) are

Many AI systems these days are adequate counterfeit humans in ways that adding machines and spreadsheets (and advanced systems that reduce to these things) are not. In other words, computers interpreting things in ways that counterfeit humans or computers generating things that counterfeit humans needs to be banned for government use. I’m not talking about banning the steam shovel but systems where it becomes difficult to determine whether or not a human did it.