Before we allow AI on the road, we'll need to have some kind of regulation on how the AI works, and who has what level of liability. This is a debate that will need to happen, and laws will need to be made. For example, if an avoidable crash occurs due to a fault in the AI, I would assume that the manufacturer would have some level of liability. It doesn't make sense to put that responsibility on a human passenger who was using the car as directed. On the other hand, if the same crash is caused by tampering performed by the owner of the car, then it seems that the owner would be liable.

As far as I know, even these simple laws don't explicitly exist yet.

Normal cars also make it easier to commit terrorist acts and other crimes. So what? I mean, yes, let's consider whether we want to take special safeguards and regulations regarding AI cars, but this shouldn't be something to go crazy worrying about.

Yes, I agree completely. I do kind of hate coming back from vacation to a huge inbox, but on the other hand, I do things like emailing someone saying, "I know you're on vacation and I don't want you to do anything now, but I know I'll forget if I don't send this now. When you get back..."

Oh, I wouldn't be surprised at all. I've worked in IT support for a couple of decades now, and I know exactly how that goes.

There are two things about that though. First, it's a bit of a fringe case. You have to consider the question, "How many people of that sort are in my target audience?" If the answer is "a lot", then you should think about writing documentation for them specifically, and find a way separate it out from other documentation for those who are more comfortable using a computer. Otherwise, people who know what they're doing are going to be frustrated searching through 100 pages of inane instructions to find actual information.

Second, people like that often also won't read the documentation. If they do, they won't understand it, or else won't feel confident that they understand it. At a certain point, you have to either provide those people with IT support personnel that they can call (your older relative has you). At the very least, you need to provide them with simple step-by-step instructions that never vary, where they don't even need to understand what they're doing. Like "In order to do [x], press the power button on your computer to turn it on (it's located in the top-right-hand corner of the box under your desk). It will flash some things on the screen for a while. Wait for it to ask for a password, and then type 'hunter2'. Wait 2 minutes. Then find the blue "E" on your screen, with "Internet Explorer" written under it. It will be the third little picture on the screen, all the way to the left..."

I've had to write instructions like that before, and some people need it to be that simple. But obviously a web application vendor can't take responsibility for that level of instruction. Even something like Dropbox, which is designed to be extremely simple, has to assume some level of competency.

Interestingly enough, while 99.9% of the world isn't in jail, quite a few of them are here in the U.S., many for "crimes" that nearly all Americans think aren't crimes. It's a sad irony that crimes that could actually cost people their lives aren't prosecuted, while "crimes" that have no victim fill our jails with "criminals."

Asking fellow geeks what chair they sit in sounds pretty on-topic to me.

I have a chair that my wife refers to as the Wall-E chair. I use it with a reading table to hold my laptop, a bluetooth keyboard, and a trackball that sits on the arm of the chair. Works a treat. The chair is something like this: http://www.la-z-boy.com/Produc...

It's ugly, the construction is pathetically bad, I've had to screw it back together with drywall screws, but it is quite comfortable, and doesn't look _completely_ awful. I'm sure you could come up with something nicer, though.

He gave us both the word "crap" and "ballcock"? What a guy.

See there, you're talking about *actually* unwanted programs. If a program is potentially unwanted, then it's not currently unwanted. It just might become unwanted in the future.

No, I mean that by any rational standard, deliberately creating a pathogen that can kill millions, whether you release it deliberately or not, is a crime. It may not be on the books, but it should be.

I think you mean something like "unwittingly installed program" then. You could knowingly install an application and still retain the potential for not-wanting it.

You are correct—this isn't attempted murder. But IMHO it's in the same moral category. I think you are basically right that it's a failure to think outside of the immediate problem space, but what a failure. Imagine if 40-60% of the people you have ever met or heard of, as well as those you don't, died within a month. The 1918 flu left emotional scars that persisted for generations. And that had a 2% mortality rate. The amount of suffering this person could have caused through his narrow thinking is more than has ever been experienced in all of history.

So if I fill a big, weak tank with a poisonous substance and deliberately park it upstream from a public reservoir, but don't actually open a valve to dump the toxin into the reservoir, and there's really only about a 20% chance of the thing bursting and dumping the whole load of toxin into the reservoir, you are saying that I have done nothing wrong, and should not be subject to prosecution, because although I set up a situation with a real probability of poisoning the water supply, I didn't actually poison the water supply.

ISTM that you are really saying that it's only attempted murder if the toxin actually winds up in the water supply; if so, you don't understand what attempted murder is. It's when you try to kill someone, but fail. I suspect that there's no mens rea here, so the charge wouldn't stick, but knowingly manufacturing a pandemic virus ought to be a crime, and the knowingly part would then constitute mens rea.

This wasn't a thought crime. This was an actual crime, knowingly committed with full buy-in from the administration. Actual virus was produces, and had the intended effect.

I know it's started becoming a common terminology, but I don't really like the terms "Potentially Unwanted Program" and "Potentially Unwanted Application". Any program/application is *potentially* unwanted. Whenever someone starts talking about PUP/PUA, I can never figure out where they're drawing the line.

I'm not aware of Odoo, so I'm only speaking generally, but generally speaking, having a simple/intuitive product does *reduce* the need to documentation. For example, I don't need documentation to tell me that I can open a Word document by going to "File", selecting "Open", and then going to "Computer", "Browse"....

Now, come to think of it, the process for even something as simple as that has gotten needlessly complicated. WTF is Microsoft doing these days?

Back on subject, yeah, if you open files by going to an obvious menu button that says, "Open File...", then I don't think you really need to document that. You only need to document the features that aren't completely blindingly obvious.

The need for documentation can also be reduced by having a good help/support system. If you have a procedure for doing something unusual and complicated that's undocumented, you had better have someone standing by that I can call/chat/email who can help me out. And even still, that stuff should be documented at least well enough that you can train your support staff.

If you don't have good support and something is not completely obvious, then yes, it should be documented.