zlacker

>>chagha+(OP)
> Machines shouldn't act like humans

I'm surprised - does that mean ML/AI isn't calm ? Why ?

>>rainha+Fd
It highlights the issue that when a machine does a task, usually the best way to do it will be not like a human would do it. Even if we had cheap full-general-AI humanoid robots to do the tasks, it may be better to do the necessary interactions as with a machine, not as with a fellow human - even in a pure business transaction much of our communication is determined by social factors, the fundamental person-to-person relationships.

For example, if I go to a dealer to buy a car, if that process would be replaced with technology, then it would be wrong to make that technology to parrot the interactions I'd usually have with a human salesman. If technology is used to take orders in a restaurant and serve food, then imitating a waiter or waitress might be a marketing novelty, but it's not the appropriate way to do it. It's a different medium now, machines shouldn't act like humans.