Transcript EECS 690
EECS 690 March 22 Some Definitions • Moral agency: The fact that something can cause harm is not enough to make it a moral agent. A moral agent must have some appreciation of morality as a concept, or should have the ability to have such a concept. • “This goes beyond traditional product safety. Of course, robots that short-circuit and cause fires are no more tolerable than toasters that do so” (p.17) Autonomy • Autonomy is a crucial concept in this section. In ethics, greater autonomy is generally seen to entail greater responsibility. • In human communities, greater autonomy tends to be earned over time. Consider what happens when the only criteria for awarding additional autonomy to artificial systems is merely instrumental. Degrees of autonomy • Many different systems are being considered here. ROV’s, for example, are not autonomous at all, so ethical questions that surround them include design issues, use issues, and abuse issues. • Semi-autonomous machines are discussed as well, but all the autonomy in these cases is instrumental. The interesting question is at what point instrumental autonomy or semi-autonomy is insufficient. Semi-autonomy • Consider this example: – “Despite NATO's efforts to keep its war against President Slobodan Milosevic as surgically "clean" as possible, the bombing of Yugoslavia has killed, ton for ton, as many civilians as other air campaigns of the past quarter-century. The weapons dropped on Yugoslavia are more accurate than those of past wars. But that very fact has emboldened commanders to drop more of them on targets that require accuracy -for example, a particular building on a downtown street. And since, as Pentagon spokesmen note, some of these bombs are bound to miss, more civilians die than anyone had predicted. "It's like engineers building bridges," said Eliot Cohen, a professor at the Johns Hopkins University's School for Advanced International Studies and chief author of the Air Force's five-volume post-war study of the Iraq air campaign. "Technology is so much better now, you'd think there wouldn't be any failures at all."So why do they happen? Because, with the new technology, engineers will go closer and closer to the tolerances of what the materials will handle," Cohen said. "They're pushed to their limits -and so, something goes wrong."” Semi-autonomy • Consider how the concerns of design and use intertwine in the preceding example. • Greater autonomy that is still only semiautonomy simply brings greater concern. • Part of this concern is in a psychological habit of persons to treat semi-autonomy as full autonomy.