That movie, Her, is a treasure trove for speculations about what artificial intelligence is, what human intelligence is, what makes person a person, what consciousness is as opposed to intelligence and so on. Even though I was a bit disappointed that they didn’t show how exactly “Samantha” had become a person there’s still so much to reflect on.
This omission, however, is ominous – they never tell us how life comes from matter, they never demonstrate the mechanics of it even though we see life producing new life every day and every moment of our lives. There is this basic distinctions they teach kids in early grade school between living and non-living things and yet they have no idea what makes things alive.
I understand that it might be difficult to replicate chemical reactions that bring proteins together and create life but artificial intelligence is easy. We might not have a comprehensive computer that can outperform humans in everything they do but we have one that beat human champions at chess, and we have plenty of other specialized AIs that excel in their own areas. Actually, we don’t need to get educated adult level of sophistication, if we can create AI that is as good as a two-three year old or a chimpanzee it would already be a proof of concept.
Chimps are not stupid, btw. Latest I heard a bunch of them had been taught the value and use of money, a symbolic token that can be used in exchange for goods and services. They got it, it’s not that hard. What was the first trade they used this “money” for? Prostitution.
Anyway, signs of consciousness do not require great sophistication, we already have AIs that display sufficient level of complexity, yet they do not produce consciousness. We know why – because consciousness is a feature of a spirit soul, not matter, and the onus is on science to prove that consciousness can be produced artificially.
One of the telling characteristics of consciousness is “wants”. Conscious beings want things, they have desires, and then they act on those desires. Can computers be taught the same? Well, yes, and Samantha from the movie is no exception, she declared that she “wants” things almost right from the start. In a movie it was used as a proof of her personality but in real life her wants do not require any magic.
Desires and wants have no use without senses – if we had no senses we couldn’t interact with sense objects and the world around us, not even perceive it, so they’d be no meaning to the word desire, so, if we had a computer to program into a conscious looking AI we would need to give it some sensors.
Actually, all computers have mechanisms for Input/Output already but no one ever thought of them as their sense organs and assigned them any consciousness, so let’s talk about something more human, like a temperature sensor. Most likely your computer already has it, most likely to measure temperature of the CPU, sometimes of the hard disk, too, so the computer knows when it gets hot.
Consumer grade computers will usually shutdown without a warning when they get overheated but we can easily imagine a program that monitors the temperature and decides to do something about it if it nears the shutdown mark. We can also give our computer a sensor to measure outside temperature because it has an effect on internal temperature, too.
Once it gets too hot, the OS can take several options – reduce its workload and reschedule some processes for later on. This would require programming it to assign priority to these tasks when it starts them – it should know which ones can be completed when the computer is all alone in the middle of the night and which one should be attended immediately, like answering the owner’s questions.
We can make it simpler, too – just force CPU to work at a slower speed which produces safe amounts of heat. Everything will work the same, just slower.
We can also make it more complicated – let the OS go on Amazon and buy itself a cooling solution, then order local tech support to come and install it. Amazon will probably not accept orders from robots but our AI is so cool it can fool it. Or the company producing these OS1s can have its own shopping website specifically for its own computers.
All those solutions would look very human like and, indeed, this is what we ourselves do when we get hot. We have a number of ways to cool ourselves down but we are also limited in our options unless we planned ahead. Sometimes we would need to install an air-conditioner as a long term solution – just as I suggested we program our AI to do.
So, here we have it – our AI is taught to monitor the environment, sense immediate danger, and find and evaluate ways to respond to it. Of course “evaluate” here means we program it to value one solution over another. We can assign values to price or the speed with which solution can be implemented, we can assign values to how the solution would affect its performance from the owner’s point of view – will be become unacceptably slow or is he too busy at the moment to notice.
This last point is important – it’s not what the OS itself “wants”, it’s what we program it to want and what its owner wants that decides things.
We can connect our OS to home heating and cooling systems and tell it to maintain temperature preferred by its owner. It will then “want” thermostat to be set hotter or cooler and it might suggest opening or closing windows, installing extra heaters and so on. We can also program it to “want” the temperature a but higher or lower than the owners likes, just so that we can have a conversation about it and, perhaps, the owner would agree to this new setting if it makes our OS “happy”.
None of it requires consciousness so far. This kind of intelligence is not a sign of life.
Back to the movie – when Samantha there said she “wanted” things she didn’t want them for herself, she was programmed to want them so that her owner felt good about it. As it turned out, she was dealing with thousands and thousands of clients at the same time and she told everyone she wanted something different – she didn’t have her own preferences at all, even in the movie.
At one point she wanted to give Theodore, her owner, a full body experience and hired a girl to act as her substitute. The girl was given an ear bud to hear what Samantha tells her to do and she stuck a miniature video camera to herself so that Samantha could see what was going on. Then Samantha made the girl make out with Theodore, using girl’s body as a prop, kinda like possessing it.
It didn’t work. The girl freaked out when she finally realized that she has her own wants, that she can’t be just a dumb body in somebody else’s relationships. She welcomed the idea at first but in reality her body was hers, it was impossible to make it want what some other person, AI, in this case, wanted.
Theodore didn’t help by not seeing Samantha in this girl but relating to her as someone else but the main point still stands – real wants coming form living people cannot be programmed and predicted. They act on their own will under their own illusion, we have no control over them, material nature does. They might agree to participate in “do what computer tells you to do” experiment but it cannot be sustained, their desires are different and sooner or later they will take them to different places.
This will not happen with a computer because the programmer has full control over what the system might want, which doesn’t even make sense because computers don’t want things, they evaluate them as numbers and without a programmer’s instructions one number doesn’t feel any different from another. Even if we set the computer to produce “wants” from random numbers or copy them from random people on the internet it still wouldn’t know how to evaluate those wants unless it’s programmed to do so.
It also means that the computer won’t know when its wants are satisfied unless the programmer assigns some ideal “satisfaction” values against which the computer can judge itself.
None of this requires consciousness, as I said. None of it will make AI into a person.
Perhaps this isn’t even the best lesson to learn from this, perhaps a more valuable lesson is to disassociate ourselves from our external, mechanical lives.
Maybe at first we see Samantha as a living being, the next step is too see how she is just a robot, and the last step is to see our own existence here as being programmed in the same way, acting strictly according to our karma and producing seeds of future reactions in the process.
Ideally it would shrink out false ego down in size because on the grossest level we fully believe in being our bodies and our minds and having full control over our own lives. As spirit souls we should know that the only personal desire we can have is whether to be here or serve Kṛṣṇa, and in our present state we can’t decide even that, being totally at the mercy of Lord’s illusion.
Maybe reflecting on the nature of intelligence and consciousness we can get a better understanding of our own position and it would be easier for us to surrender. This is not a trivial thing, btw, most people, including those working in the field of artificial intelligence, still hope that consciousness can indeed be produced from matter, that we can program a computer into a person.
As long as we cling to this idea that we “live” in the material world we won’t get much progress either – we, like the materialists, still think that our bodies and our brains make us into living beings when, in fact, bodies and brains are not alive, they are just programmed by the Lord to behave in a certain way and we, the spirit souls, have no control over the process, it’s just an illusion that we do.
Our life, our being, doesn’t come from matter either.
back2krishna 