How can civilization handle people being able to produce disposable intelligent autonomous agents?

Question

In the novel Who Censored Roger Rabbit?, rather than the 'toons being indestructible as in the movie, they can produce doppelgängers that are a mere shell of the being, which fade out after a few hours. One character commits a murder and uses a special high-quality doppelgänger to establish an alibi. Now this begs the question of why not have the dop commit the crime and then run off to expire, so it can't be tied back to him at all?

In David Brin's magnificent Kiln People, people make golums of various fidelity to do chores or even work; some have factory workers generated daily to expire rather than re-merge at the end of the day.

In a Larry Niven story, the character had his memory restored to an earlier backup so he was unaware of his escapades.

In many stories, we have robots, usually programmed with safety features so you can't just order one to rob a bank or kill someone.

If people could create disposable copies, or order about robots complicated enough to pull off a bank robbery, it would have the general property of allowing people to get away with things.

How could society prevent this misuse, or cope with such reality?

Answer 1

The same way that it currently copes with someone hiring an autonomous individual to commit crimes for them.

If physical evidence and witnesses are non-existent then ability, motive, and opportunity all play a role in determining the ultimate source of the crime.

Answer 2

I suggest the approach is simple - collective responsibility. If one dop commits a capital crime, all instances, including the original, are executed. This neatly avoids the determination of which aspect is guilty and which is not.

For non-capital crimes the situation becomes murkier, but collective responsibility will still serve. Property crimes such as theft result in a fine equal to some multiple of the stolen property value, and is levied on each dop. So, if you (dop #3 of 10) steal 1000 dollars, each dop gets fined (let's say) 3000 dollars. 3000 goes to the victim as restitution, and the rest goes to the state.

Bodily harm cases are difficult, and I can't think of how to apply the principle, so I'm open to suggestion. Perhaps the victim is permitted to inflict a similar level of harm on one of the dops, selection to be made at random.

Answer 3

Since the question is how to prevent criminal or unethical activities by the disposable avatars, then the first step would be to secure the facilities used to make them.

This does not sound like "kitchen science" anyway, but with a secure facility, you (instance 0) will need to register when you visit, supplying ID, DNA sample and a non refundable deposit. Any activities from the avatars will eventually have to be traced back to the secure facility, where the records are kept.

Step two would be to impress some sort of serial number on the avatars. I'm going to imagine the avatars are a form of clone, so the DNA can have a special sequence coded into the "junk" DNA which basically says "Copy 1 of John Doe" (insert number as appropriate). The nature of the serial number will depend on how the avatar is actually made, a checksum embedded in the downloaded personality, a physical serial number on a robot, but there will always be some sort of means of determining the nature of the avatar (the idea the avatars completely vanish at the end of the time period has some issues, and I suspect that there will always be a physical shell of some sort to examine).

Step 3 would be a separate records backup at another secure facility. If avatar THX1138 is accused of a crime but the secure facility and secure records all agree that only 1137 instances were made, then you (instance 0) have an alibi; this was down by an unauthorized copy.

Even these steps are not foolproof, since autonomous agents will by definition be able to work and plan on their own. THX1138 could have planed the robbery on its own, or committed the murder as a crime of passion when it was jilted. In these instances, normal police work will be needed to determine "who" or "which one" was present, and then the usual Method, Motive and Opportunity applies. Proving a conspiracy will be difficult, but in a society like this, it would be very stupid for instance 0 to not have their cellphone set to record every conversation and text at all times, just to establish your own innocence.