Empathy is something that evolved in animals under very specific conditions. And even then, about 4% of people are psychopaths or sociopaths with little or no ability to empathize with others or feel guilt. They haven't been removed from the gene pool because a lot of the time that is an advantage.

We should ideally try to create machines with empathy and morality like humans, but that isn't a simple problem at all. And if the first AIs are created without it, they will essentially by psychopathic monsters with far more intelligence than the best human psychopaths. This could easily go very very bad, which is why we are trying to figure out how to create AIs with morality and why people are taking the problem seriously.

Giving someone much longer time to think about something is still a huge advantage. Besides an AI won't necessarily share the same biases that humans have, and it will probably be capable of modifying it's own code to remove them and other inefficiencies. A human engineer with millions of years to work on something (assuming they could modify themselves enough so they don't get bored or go insane) would be a very scary thing, even with a normal human mind.

Computers are almost certainly going to become more intelligent than humans at some point. Probably sooner rather than later if you look at how fast technology has advanced in the past and is currently advancing. And certainly a machine much smarter than humans would be a danger to us if it's goals were not exactly the same as ours. These "snobs" are some of the only people in the world taking a potential world ending risk seriously and trying to do anything about it.

At one time flying machines that could rain down explosives on cities would have sounded completely insane. Let alone nuclear missiles that could launched anywhere in the world and kill millions instantly. This was the kind of thing that science fiction writers of the time wrote about, but few actually believed or considered seriously. It sounded absurd, and absurd sounding things never happen, right?

Fortunately those things only killed millions, not billions (though they easily could have). But we can't wait till the last minute to start taking AI seriously. If the first AIs are created with goals that are different than humanities, but are much smarter than us, there is no telling what kind of damage they could do. We may only get one chance to not screw this up. Please don't throw it away.

Machines made by humans still haven't beaten machines made by 4 billion years of evolution in *all* aspects. That doesn't mean we could never figure out how, let alone something that is smarter than us couldn't figure out how. We are certainly smarter than evolution. We don't try random changes and do real world tests millions of times in order to engineer even simple things. We can definitely do better than evolution, it just had a head start.

Besides it doesn't really matter. A horse might be able to use less energy and repair itself much better than a car, but if you want to get somewhere fast, you are still going to use the car. It may not have all the advantages of a horse, but for what we want it to do, that is travel places, it is simply superior. Likewise we could design a brain that doesn't have all the advantages of a human brain, like the ability to self-repair or take up little space and energy, but is still smarter than us.

We've already done this actually, computers are far smarter than humans at a lot of things. Literally millions of times smarter at some things. Someday we may make one that is better than us at the other things as well.

It's not trivial to kill another human being. Medically yes, we have bodies that are not designed to withstand bullets, knives, or radiation. But if you actually want to kill someone, they can fight back. They can design weapons. They can run or hide. They can try to convince you not to kill them or convince others to help them.

It's important to understand that an AI isn't just going to be a fancy computer in a room somewhere. If it's connected to the internet it can upload itself or other AIs to computers all over the world. And it probably won't start a war with humanity before it has a physical body somewhere safe that we can't find or easily destroy. Assuming there's a war at all and not something more alike to a person stepping on an ant hill.

It doesn't have to be directly hooked up to those systems. If it's smarter than humans, maybe even many times smarter, it might find flaws in the security systems better than human hackers, guess passwords, dig up information on people and trick or blackmail them, etc. Yes this is all speculation, but a being many times more intelligent than you is probably going to think of all this as well as ideas we can't even imagine. Even if it fails the first time, people are going to keep building AIs and computers are only going to get faster and faster and more and more connected.

Human minds are pretty intelligent, and we don't have any of those problems. We require relatively little energy, resources, and space. Imagine desinging one much better (evolution is extremely inefficient, it simply had more time) and many, many times bigger and with all the advantages modern computers have on top of it. If you can think of a way to destroy it, it can think of it too and, being many times smarter than you, come up with a way around it. Hiding, shielding itself, designing a virus that kills humans, whatever. This is just speculation of course, I have absolutely no idea what a being many times smarter than me would do, but I wouldn't want to be in it's way.