Robotics Researchers Attempt To Model Morality
Writing by Evan Ackerman on Friday, 18 of December , 2009 at 12:27 am
You’re probably familiar with the Trolley Problem ethical thought experiment, but if you’re not, here’s the question:
A trolley is running out of control down a track. In its path are 5 people who have been tied to the track by the mad philosopher. Fortunately, you can flip a switch, which will lead the trolley down a different track to safety. Unfortunately, there is a single person tied to that track. Should you flip the switch?
There are a whole bunch of different variations on this scenario, involving fat people and organ donors, but according to morality and ethics researchers, “factors such as gender, age, education level, and cultural background have little influence on the judgments people make, in part because those judgments are generated by an unconscious “moral grammar” that is analogous in some respects to the unconscious linguistic grammars that support ordinary language use.” So basically, there are ‘built in’ rules of ethics that humans have.
Lu├¡s Moniz Pereira of the Universidade Nova de Lisboa in Portugal and Ari Saptawijaya of the Universitas Indonesia have published a paper entitled “Modeling Morality with Prospective Logic” in which they discuss using the answers to ethical questions like the Trolley Problem to create a piece of software that answers such questions the same way a human would. This is not yet a broad spectrum ethical governor like Ronald Arkin is working on, and it’s not designed for military situations, but it does suggest that it’s possible to endow robots with software that gives them the capability to make ethical decisions that are at least consistent with the decisions that a human would make. Whether those decisions are best, of course, is an entirely different matter… But that’s the great thing about robots: when you find something that works better, you just reprogram them.
VIA [ Technovelgy ]
Comments (5)
Category: Artificial Intelligence,Research
- Add this post to
- Del.icio.us -
- Digg
Comment by moron4hire
Made Friday, 18 of December , 2009 at 5:55 am
“…using the answers to ethical questions like the Trolley Problem to create a piece of software that answers such questions the same way a human would…”
This sounds an awful lot like an “Expert System”, which barely classifies as AI. It’s just taking a set of known answers to a set of known questions and parroting them out at the right time. Expert systems have been around for decades, they’re nothing spectacular.
Comment by Evan Ackerman
Made Friday, 18 of December , 2009 at 6:16 am
I think the difference here would be that the software is not designed to answer known questions with known answers. Rather, by analyzing known questions and known answers, the software has the ability to apply the underlying “moral rules” to unknown questions and still provide answers that are consistent with the answers that a human would give.
Comment by Barrett Ames
Made Friday, 18 of December , 2009 at 10:10 am
It’s more like a classifier than an expert system the way Evan puts it.
Comment by quantum_flux
Made Saturday, 23 of October , 2010 at 8:23 pm
How can you model something that isn’t mathematized? OTRAUMP!, wait a second, it is mathematized.
Comment by Lyndsay Willmarth
Made Wednesday, 14 of December , 2011 at 10:56 pm
Great content material and great layout. Your blog deserves all of the positive feedback it has been getting.
