This site may earn chapter commissions from the links on this folio. Terms of use.

robot reading

I of the more disturbing trends in robotics is how often some researchers gloss over the moral complexities of AI past suggesting Isaac Asimov's "Three Laws of Robotics" will be sufficient to handle any ambiguities robots come across. This only serves to demonstrate how unfamiliar many technologists are with the depth of the upshot, for as a closer reading of Asimov's work reveals, the iii laws of robotics leave plenty of room for disastrous outcomes.

At that place are encouraging signs, even so, that at least some other researchers are taking the trouble seriously. Two figures leading the accuse in this direction are Mark Riedl and Brent Harrison of Georgia Constitute of Applied science. They are pioneering a system chosen Quixote, by which an artificial intelligence learns "value alignment" past reading stories from different cultures.

This may seem a strange approach to the problem — until 1 considers how regional questions of morality are. For example, two counties only a few miles apart in the United States may take vastly unlike ways of perceiving moral and upstanding problems. For instance, spanking your child may be considered a morally unacceptable act amongst an flush urban population, whereas immigrants or more rural communities may find this grade of punishment more acceptable. This is non to degree aspersions in either management, but just to indicate out that even among humans, there is petty consensus on moral and ethical issues. How much more difficult will it be for humans to concord on what constitutes upstanding robot behavior?

Quixote system

The Quixote arrangement, by which an AI learns ethical behavior through reinforcement learning and multicultural stories

As information technology turns out, there may be a way, and it'due south not so different from how humans learn the values particular to the region they grow upwards in: by telling stories to each other. The fables and legends item to a locale are oftentimes laced with moral directives that provide clues to a growing child on what constitutes morally acceptable behavior in their culture. The idea behind Quixote is that the same principles can exist applied to shaping robot behavior.

Quixote builds on Riedl's prior inquiry project, which demonstrated that an artificial intelligence tin can exist trained to identify a correct sequence of actions by crowdsourcing story plots from the Internet. This is not at all dissimilar from the RoboWatch project nosotros reported on earlier in the year, which showed how an AI could be trained to learn simple household tasks like the steps involved in making a peanut butter sandwich by watching YouTube videos. The Quixote organization takes this a step further, adding a reward signal that reinforces certain behaviors and punishes others during a trial-and-error learning procedure. In essence, the AI learns to imitate the beliefs of the "good guy" in the stories it reads from the Net and avert the beliefs perpetrated by the ignoble characters.

All the same, the arroyo is non without its pitfalls. In some parts of the world, for case, female person genital mutilation is considered de rigueur, though the majority of humanity regards such practices as objectionable and would certainly not want robots perpetuating them. In other words, in training robots on stories modeled later on our own cultural precedents, we may air current up giving new life to behaviors that are improve left to wither on the vine. Therefore, formulating a robust and universal upstanding organisation — usually the pastime of ivory tower philosophers — is likely to have on increased significance equally we consider the possible repercussions of endowing robots with the same porous and inconsistent moral norms exemplified by our own regional populations.