Possible applications of Mīmāṃsā deontics: on Chaudhuri and Vardi

There are fields in which the contribution of applied ethics and deontics are more than needed, such as that of the programming of artificial intelligence connected to robots which might interact with human beings. Chaudhuri and Vardi (their article can be downloaded here) quote the following case:

Once released on our streets, self-driving cars will almost certainly face circumstances where they just cannot avoid harming a human.
It is thus imperative that the new generations of intelligent machines be not just efficient or functional but also ethical.

A case which might appear even closer to readers of this blog is that of softwares needing to handle the allocation of resources. These need to receive several bits of information, but to neglect some of them (e.g., “membership in a political party”), while taking into account others (e.g. “academic ability”, which Chaudhuri and Vardi do not explain further).
Now, all these issues do not only need the help of ethicists who are willing to engage in the elaboration of appropriate programs, but also of deontic logicians. Consider the following example, again from the same article:

[W]hat if a drone in a battlefield is obligated to ensure p, but it is impossible to ensure p? Is the drone still required to fulfill its other obligations? The answer should presumably be yes. However, note that this is a departure from classical [deontic, EF] logic, where the impossibility to meet p would amount to an inconsistency, rendering moot every other consideration.

This entails the need to elaborate a deontic logic which is no longer bound to the limits of Standard Deontic Logic and, for instance, values differently different obligations, so that a failure to accomplish a minor obligation does not make the system collapse. This consideration would already be enough to make researchers curious about deontic systems developed outside the Lutheran tradition (it might be worth remembering that deontic logic started in the West in Scandinavia and has until now been primarily developed in Northern Europe and the US) and thus entailing, most probably, different biases, so that the combination of various systems could pave the way for the creation of a more comprehensive one, able to process different sorts of conflicts. It is in this connection perhaps not surprising that the authors of this article are both coming from outside the precincts of the majority of scholars working on deontic logic (respectively, from India and Israel).
On top of that, there are specific reasons to want to come closer to the Mīmāṃsā deontics, since it precisely entails a hierarchy of prescriptions/obligations (see the comments to this post).

Now, a different question: On which side does it depend that programmers and logicians interact so little with historians of philosophy?

For another post on Mīmāṃsā deontic read by contemporary logicians, see here.

(cross-posted on my personal blog)

About elisa freschi

My long-term program is to make "Indian Philosophy" part of "Philosophy". You can follow me also on my personal blog: elisafreschi.com, on Academia, on Amazon, etc.

2 Replies to “Possible applications of Mīmāṃsā deontics: on Chaudhuri and Vardi”

  1. Nice post, Elisa. I wrote about the same topic a while ago from a slightly different tack: http://loveofallwisdom.com/blog/2012/06/on-the-ethics-of-robots/

    The deontic logicians seem to be following Kant’s assumption that “ought implies can”. I presume that in such a system the response would be: if it is impossible for a drone to ensure p, the drone is then ipso facto immediately no longer obligated to ensure p.

    (Now you’ve got me thinking about designing an analytical ethics simulator program – design AIs in a virtual space that you can program to act according to Kant, Mill and Rawls…)

    • The last point was in fact the idea… but I guess that one also needs a robust logic to tell the drone (etc.) what to do in a landscape which cannot be completely foreseen.

Leave a Reply to elisa freschi Cancel reply

Your email address will not be published. Required fields are marked *

*