Often it would be nice to be in two places at once. There is a conference you want to go to, but there is also work to do at the office. There are Christmas vacations on different sides of the family, and you want to be at all of them.
Suppose that in the future you can duplicate your consciousness and imprint it onto a synthetic robot that is sitting inside your closet. You activate the robot, duplicate your consciousness onto it, and then send the robot off to the conference to attend on your behalf.
When the robot returns, you look at the experiences it has had, and decide whether you want to merge its experiences back in with yours. If you merge, you would then remember the event from a first-person’s perspective. If there are any conflicts in the merge, e.g. new beliefs that the robot has acquired while being away, you would resolve the conflict to keep the beliefs that you wanted to keep.
You then wipe the consciousness from the robot, turn it off, and put it back in the closet. Over time you can imagine the robots getting more and more life-like, until they look like humans, and do a good job representing you.
Ethical challenges
Even if this technology was possible, there would be an ethical challenge to its going mainstream. Once you put a duplicate conscious state into the robot, the robot is now a conscious agent, and as deserving of moral consideration as you are. If you turned the robot off while it was protesting and saying it wanted to live longer, it seems that this would satisfy a reasonable criterion of murder. It would be natural for robot rights groups to form that protested against the treatment of conscious robots, pointing out that if an agent is conscious, it shouldn’t matter whether the agent is made of skin and bone, or something else.
Is there a way to get around this ethical roadblock? One tactic would be to include in the copy your conscious states in robot the desire to be terminated.
There would be two problems with this. One is that robots rights group would meet these robots at conferences or wherever they are and point that they are effectively brain-washed, and offer to delete that self-terminating desire from their conscious state. When you sent your robot off to act on your behalf, you wouldn’t know whether they would come back and still have that self-termination desire.
The second problem is that it is not clear whether it’s ethical to include a self-terminating desire in the robot’s consciousness. Suppose we could tweak DNA and someone created a child that wanted to self-terminate after 5 years. I imagine there would be an ethical reaction to that.
When I was chatting with John Hawthorne, he came up with an interesting idea. What if the robot was just streaming consciousness to you via the cellphone networks, and the actual physical seat of consciousness was in your brain? In this scenario your brain is the seat of consciousness, and the robot is a far-away input to your conscious system, and it’s not itself conscious.
Let’s suppose that the physical seat of consciousness was a chip in your brain: this chip did the thinking and experiencing and had the conscious states. Then there will be ethical implications of switching off this chip when you want to turn the robot off. The chip is a conscious agent with its own desires, and it has a right to life in the same way that you do.
What if you never switched it off, and you just activated the robot when you wanted? In this scenario, the physical chip is always conscious. Sometimes it is connected to an external robot, and sometimes it is disconnected, and it has to make do with its own thoughts.
One issue is that if the chip never turned off, and was left to wander with its own thoughts, it would develop its own personality, distinct from the personality of your brain. Then it would no longer be as effective, if the idea is that the robot can represent you at conferences and other places. Wiping the consciousness of the chip every time you wanted to activate the robot would be considered unethical, just as unethical as it would be to wipe the consciousness of a human agent without their consent.
There may also be ethical implications about activating and de-activating the robot, which is the chip’s access to the world. It would be equivalent to periodically inducing in a human a state of not being able to experience the world or touch things, and then periodically re-activating that access.
What if you could do this forking of consciousness without any new hardware in your brain, no special chip? What if you could spin up a kind of sub-routine in your brain that was a separate conscious process, and that could act as the basis of streaming for the robot?
I don’t think the hardware is the main issue here. The complication emerges when we consider whether the conscious sub-routine can form separate desires from the master conscious process. Suppose it can, and that the sub-routine forms the desire not to be shut down at the end of the conference. Against this desire, the master conscious process shuts it down. It seems that this is analogous to the conscious chip situation. The sub-routine has as much right to survive as the master process, as they are both conscious processes.
Could you create a conscious sub-routine that couldn’t form its own desires? This seems hard to imagine. You want the sub-routine to be able to form desires such as “I want to catch this plane”, “I want to talk to that person”. The sub-routine could be set up so that it consults the master conscious process to formulate desires, but then the master conscious process would be bombarded with desire requests every second or two about what desire the sub-routine should formulate next, and this would remove the benefit of forking your consciousness in the first place.
In short, even if we do manage to figure out how to fork consciousness and imprint it onto duplicates, it seems that once you spawn a conscious agent or process, you have spawned something that deserves moral consideration as much as you do, and so switching the conscious processes off is not ethically permissible.