This statement will have zero impact on subsequent sensational headlines or posters here claiming Google is making killbots.
- Is pursuing an AI that kills 10 bad guys but saves 20 good ones overall bad?
- Is pursuing an AI that doesn't kill anyone but pushes us to watch ads and lose our life on youtube overall bad?
Not commenting on whether that's accurate; just trying to better articulate what I believe is the above poster's philosophical argument.
- Depends : How humans choose to spend their lives is their own issue. Personal responsibility dictates it is each individuals choice how to live their lives. Youtube is not some resource needed for life.
Either way this is one of the reasons I'm not wild about the crazy self-driving race.
If it senses an unavoidable collusion ahead, it will need to choose from possible reactions that each have different consequences for the various people and property. Someone has to program or train the car what to value when choosing its reaction. The car will be biased, implicitly or explicitly, by the programmer's values.
Alternatively, you can have the people vote on a self-driving car referendum, which specifies a DMV-enforced policy for the trolley problem.
As a society we vote on such matters all the time. Nothing new here.
MIT created an online survey called the Moral Machine asking people to choose who a car should hit in various (randomized and user-created) scenarios. Data like that could be used to train a "fair" AI. It's an interesting thought experiment but the survey's scenarios are unrealistic and include information that a car AI would not have available, e.g. person A is homeless and has no children while person B is a wealthy executive in good health.
The trolley problem is unsolvable; that's the point.
I'm not saying the decision will be better when a human makes it, I'm saying it will be more humanely considered.
- Yes. If your philosophical development is to the point where there are clearly defined good guys and bad guys in your mind I suggest you read more.
Yes. That's kind of the problem.
Declaring that a person wearing a suicide vest in an Arab market is inherently bad does play on racial prejudice and emotional appeal. It's impossible to separate that statement from a decade and a half of propaganda, and the trauma caused by the numerous suicide attacks on civilians in the western.
In what contexts is a person wearing a suicide vest in a market anything other than inherently bad?
Killing "bad people" to save "good people" is not a self defense argument. The statement is too general. Is it okay to do medical testing on criminals if it speeds the advancement of medicines that saves law abiding citizens lives? We could have limits to only use people found guilty of the worst crimes and increase the legal burden of proof for criminals going into the medical testing program. It would decrease overall harm. It would kill bad guys to save good guys. Still, this practice is illegal for good reason. We need a better reason to kill someone than they are bad.
[0] https://www.brookings.edu/blog/markaz/2015/12/09/what-americ...
The argument here isn't that the people have the right to defend themselves. The argument here is that a person wearing a suicide vest is bad and deserves to die for being bad. They are evil. This argument is used by governments and religions throughout history to justify killing. Often by both sides of a conflict. It's a worn out argument, old and tired. Who is the judge of what is evil? The only long standing definition for evil involves religious context to believe. Anything else is poorly defined, not widely agreed upon, or hard to test against.
'Evil-skeptics give three main reasons to abandon the concept of evil: (1) the concept of evil involves unwarranted metaphysical commitments to dark spirits, the supernatural, or the devil; (2) the concept of evil is useless because it lacks explanatory power; and (3) the concept of evil can be harmful or dangerous when used in moral, political, and legal contexts, and so, it should not be used in those contexts, if at all.'[0]
Google did not promise to employ utilitarian ideals when deciding if the outcome of a new contract or product is good or bad. They said, "We will not directly aid in killing or harming people."
However, I suspect this vow will invite challenges that question the redistribution of personal data Google aggressively collects to other parties, both advertisers and Orwellian governments. But I'm glad they spoke out nonetheless, since it invites closer scrutiny and upbraiding of the company by eagle-eyed social critics, who are all too often silent or ignored on these issues.
It's irrelevant. The swordsmith doesn't get to choose who gets cut down.
Does anyone think it's possible to build autonomous weapons that can't be turned upon arbitrarily chosen populations?