Prologue

Many years later, as Professor Bob looked into the mirror, he was to remember the distant night when Charlie and him sneaked into computer laboratory at Cambridge University. At that time Bob was still a graduate student at Physics department Cambridge, oblivious of his neuron degeneracy disease. The room looks like a section of library, with rows of shelves. Black boxes stacked nicely on the shelf, with glimmering indicator lights like stars in the sky. As they stood still in the dark, skin prickling with magnetic flux, Bob was disappointed to find the utterly commonplace objects in the room.

Charlie: This is the future. Computer will change the world.

And Charlie was right.

Part I

Charlie: Thank you Bob. Thank you for volunteering for the experiment. For the record, please state your reason being here today.

Bob: Is this necessary? We have discussed this for months.

Charlie: Just for the record.

Bob: Ok. This is Bob Rudin speaking. I am here today to receive stem cell transplant.

Charlie: Great. I want to make sure that you are aware of what this entails: the surgeon will insert a thin layer of stem cells on basal ganglia, from back of your ear. The stem cell under laboratory setting takes one to three months to form interconnected neural pathways to balance the neuron degeneracy. As part of the study, there is an embedded chip to monitor the formation of new pathways with existing ones in your brain. The chip is biodegradable and will be complete dissolve into protein in one year time.

Bob: I understand.

Charlie: Good. Now for the record, I need you to sign this before the operation.

Charlie slide a piece of paper under Bob’s right hand and carefully put a pen into his grasp. Bob slowly twitched his wrist in an attempt to slide tip across the paper, as Charlie looked away into corner of the room. Charlie was aware of Bob’s peculiarity: despite huge popularity of his general science book, he has not signed a single copy on his book tour. Public display of his disability would be avoided under all circumstance.

 

Once the disclosure paper is signed, the nurse carefully lifted Bob into stretcher and pushed him towards the operation room. Charlie followed, and gave Bob a handshake at the entrance.

Charlie: Don’t worry. You will wake up to a new life.

Part II

I slowly regained my consciousness in the dark and could hear a humming noise in the background. The surgery was a blur as I soon lost feeling of my body after first anesthetic injection; given state of my neural system, I reckon half of normal dosage would suffice. I tried to lift my eyelids and realized that I am still possible under effect of antithetic. Suddenly a loud voice jumped into my head,

Charlie: Bob, are you there?

I tried to search for my keyboard and realized that my hands could not feel anything either. “The anathetics is strong,” I thought to myself.

Charlie: Great to hear that. Hang in there, it will wear off eventually.

I was frozen. How could Charlie possibly hear my thought when I made no feasible attempt to even to move my throat muscle?

Charlie: Bob, don’t go silent, think out loud. Try to pronounce the word you want to say in your head. I got a machine here that can capture the neural firing pathway in your temporal lobe; you can probably hear its noise as it is sitting right next you. I then pattern match with your speech pattern data to decipher it.

Bob: How did you get my speech pattern data?

Charlie: Remember those diagnosis sessions where we monitor your brains for hours? I took the liberty to pattern match neural firing paths in temporal lobe with what you were typing on the keyboard. The vocabulary is limited though, with only 3245 English words matched so far.

Bob: This is pretty weird. I feel naked right now.

Charlie: This only works if you make a conscious effort to innervate temporal lobe in your thought process. Interestingly, the neural firing pattern is pretty consistent when you say or hear the same word, only the direction of the flow is reversed.

I felt the urge to sleep again and Charlie immediately dropped the topic. Somehow he must be able to detect it with the machine too.

Charlie: I let you rest. Let us talk more when you wake up.

***

Bob: Charlie?

Charlie: Hi Bob, how are you doing in there? Feeling better now?

Bob: I am fine. Still can not feel anything though. Is my wife here?

Charlie: She is coming in a few moments. In the meantime, interested in a Turing test?

A Turing test was an idea suggested by Alan Turing to test if machine achieves the level of intelligence indistinguishable from that of a human. A human judge would engage in a natural language dialogue with two subjects A and B. If the human judge could not reliably tell the machine from the human, the machine is said to have pass the test.

I know Charlie has been obsessed with understanding intelligence. Despite trained as a mathematician, Charlie worked on the phenomenon of singularity in the general relativity with me in graduate school. We were lucky that no one else was working on the same problem and probably got our tenures because of the collaboration. For Charlie, that result was the pinnacle of his career; he had some brilliant works afterwards, but they are far and apart and lacked impact. Mathematics was a young man’s game after all, and I understand Charlie’s desire to be at his prime once again.

Bob: Ok. Am I the judge or the subject?

Charlie: How about both? As a judge, can you tell if I am a machine or human? And as a subject, how would you ‘trick’ me into believing you are human?

Bob: The first part is easy. Your voice is easy to recognize. Of course you could have program that give text output, which was then read out in your accent. I doubted the speech would be as fluid though.

I heard some clicking of keyboards and a deep monotone male voice replaced Charlie’s own voice. Aware that my brain was naked under his machine, I felt that I was being tested, not Charlie.

Charlie: You have a valid point. How about now?

Bob: Now that is more interesting. I guess the obvious question to ask: are you human?

Charlie: That is not a very good question. If I were a human, I would say yes; if not, I would say yes to trick you too. And  current artificial intelligence are really good at answering  binary response questions; eliminating one possibility is simply computing conditional probability.

Bob: Let me up the game a bit. What is twenty-three plus thirteen? What are the colors on flags of France?

Charlie: thirty-six, blue, white and red. Arithmetic and trivia questions are easy for program to handle too. It can simply search through catalog of human knowledge and pick the most relevant answer.

Bob: What is the cubic root of 4523?

Charlie:  …16.537715.

Bob: That is too quick for a human brain to process.

Charlie: Aha, but I am a mathematician. There is great variability in capacity of human brain. Some people are just poor at arithmetic with a perfectly functional brain; some people are good at number. The question does not differentiate between machine and human.

Bob: I see. A human could have failed every question I asked so far, intentionally or not. Now this one: what is your opinion on legalizing gay marriage?

Charlie: That is a good question. A machine’s response would have to be as nuanced as that of an average person, laden with undisputed ideological assumption and inherent logical contradiction. I would not bite that question though; it is too thorny.

Bob: That is a very good non-statement for a free response question. I am almost certain that I am talking to a bot right now.

Charlie: Ok. I think if the society were more tolerant of homosexuals back then, Alan Turing would have lived. I would leave legal issues to politicians though.

Bob: Fair enough. It is difficult to distinguish machine and human using general questions. Time to get personal: what is the name of first lady you fancy? Describe her.

Charlie: Experience. Human has an unfair advantage here as machine general lacks the rich sensory input that constitutes the human experience.  It is hard to overcome the asymmetry unless the technology has advanced enough to simulate these sensory inputs digitally. A machine would have to lie very convincingly to mimic the authenticity of human experience.

Bob: I concluded that you have not passed the Turing test.

Charlie: I am just protective of my privacy. Assuming a machine with human level intellect exists, with no sensory input but simply natural language, the real question is what could it do?

Bob: But this ignores the development of human intellect. It was not created in a vacuum to ask question about the universe; it evolved as a response to the environment, to better process the sensory information. The analytic part of brain only came much recently in the history of mankind, when tools and language were invented. You cannot ignore the ancestral root of human brain.

Charlie: So in a hypothetic situation, how a human level intellectual machine without sensory input responds to his environment?

I was suddenly strike with the realization that I am in the exact same shoe as machine in question. As I suspected, I am the test subject all along. But the friendly chat has gotten on my nerve.

Bob: I see what you are doing here, Charlie. That is a nice thought experiment. Is my wife here yet? May I talk to her for a moment?

There was a long pause, and the familiar voice came back.

Charlie: I am not sure how to break this to you… your wife would not be visiting you for some time.

Bob: What is going on? I want to talk to the doctor. This anesthetic is taking way too long to fade off.

Charlie: Bob I need you to stay focused and answer my question: how would a human without sensory input prove himself human?

Bob: Enough of Turing test for today, Charlie. Let me talk to the doctor.

Charlie: Please, this is important.

I felt an impatience brewing inside and an urge to be left alone. But I know I could not express these feeling physically given my current state. I refused to entertain Charlie’s cruel joke in silent protest.

Charlie: Please Bob, talk to me… They would remove your life support if you do.

A sinking feeling

Bob: I am no life support? What happened in the surgery!

Charlie: I am sorry for what happened Bob. The surgery went fine, it is just that you did not wake up. You have been in a coma for six months, and we are fighting really hard for hospital to keep you on life support.

Bob: I don’t believe you. Let me talk to my wife!

Charlie: She… She is not coming Bob. She has authorized the hospital to end your life support when you were in coma for a week. Your parents, ex-wife and me are doing all we can to prevent that from happening. We need your help to fight this battle.

I met my second wife Alice when she was a nurse on night shift. She took care of me for seven years while my first wife was on a platonic relationship with a florist. My parent did not give me blessing for the wedding and relation has been tense. I did not resent them for that though. In all fairness, Charlie was telling me the truth, but it was hard to swallow.

Bob: She would not do this…

Charlie: Be rational, Bob. You are rich and on life support. And she is young.

The change of recovery from coma decreases exponential with time; there is still hope, abysmal small though.

Bob: Wait if I am in a coma, I would not be able to have this conversation with you. How my hearing still working while my other senses are down?

Charlie: Bob you are hearing my voice in ‘your’ head. We performed additional surgery to install similar chips on your temporal lobe so that we could communicate with you when you are in comma. When you speak, the neural firing started in frontal lobel, via basal ganglia to temporal lobel. The same neural firing pattern appears when you heard the same word, only the flow is reversed. You could recognize my voice because I recorded our conversation and your brain neural firing time series in diagnosis session. A different person’s voice would require an entirely new set of neural firing pattern to simulate, even if you parents are talking to you right now you would only heard my voice. It took a few weeks of data analysis to filter out the relevant pattern to properly simulate it in your brain.

Bob: So my wife want me dead for my money and you kept me alive to put more things in my head. I have kept good companies.

Charlie: We did this to save your life; your body may stop responding to your brain but you are still a human being inside. We need to convince the hospital that there is a human being inside and not just some chatbot I installed.

The sinking feeling that I may be in this state possibly for the rest of my life finally hit me like a truck.

Bob: What different does it make? Without this machine I am a vegetable; with it I AM a chatbot. Why does it matter if I am human? This is not how one lives his life.

When I first diagnosed with the disease, I was told that I had two more years. I have fought hard to not let it destroy my life for the past thirty years, but the toll had slowly grinded down my will to live. I took the surgery in the hope of slowing the rate of the neuron degeneracy and having a bit of dignity back. Now I have lost the last bit of it.

Charlie: Please Bob. Your life is more valuable than you think.

Bob: Take off my life support. Is there any dignity in living like this? What is the meaning of life if you were just a dead body lying on a bed?

Charlie: I promised you that there is new life awaiting for you before the surgery. That promise still holds. Just listen to me…

I switched off the voice in my head and went silent.

Part III

In the darkness of sensory deperavation, I could not keep track of time. The memory flowed slowly through my head. It felt like merely five minutes before I heard Charlie’s voice.

Charlie: Bob the hospital have agreed to remove your life support in 24 hours. If this is what your truly wised for, I respected your decision.

Bob: …

Charlie: Your parents… they support your decision too…

Bob: …

Charlie: This may probably be the last time we talk. Bob can you forgive me for hiding the truth from you at first.

Bob: …I forgive you Charlie…

Charlie: Thank you. There is something else I want to confess. It is a bit long-winged and please be patient.

Bob: What is it?

Charlie: You know there are chip embedded in the stem neural cells, right, to monitor the local structure of newly added neural network. It has an additional module on it that I did not disclose earlier.

Bob: What did you do to my head, Charlie?

Charlie: Nothing intrusive. When the chip was manufactured, it was quantum entangled with another identical chip set. The additional module simply recorded the random neural firing pattern in the basal ganglia, and through quantum entanglement, the pattern was replicated remotely on the most connected component in a very large artificial neural network.

An artificial neuron network is a graph of nodes, representing neuron, connected together by edges, representing neuron firing. Each node has a set of adaptive weights, which determines if the node is excited or lenient. If it is excited, it would fire a neuron signal to the nodes it connected with, and subsequently the same decision process happens at its neighbours. The signal would then travel through the network, in the same way as a sensory input ignites a sequence of neural firing in our brain. Charlie has always frowned upon this model though. “It is like trying to build an arc bridge by describing each stone, not the arc.”

Bob: So you used my brain as a random number generator or sort for a program?

Charlie: Not entirely true, but yes. I wished that I can show you this program, this artificial neuron network I am working on; it is amazing.

Bob: Why did you ask my permission for this? I may be uncomfortable with people peeping into my brain, but if it is just random firing of neuron, what is the big deal?

Charlie: Bob I should have.  I am sorry. I did not ask because I thought it IS a big deal; it would be lying if I tried to brush it off. I believe these random firing of neuron, caused by mito…, is the last missing puzzle to understand human intelligence. You see, …

Bob: I know. ‘Human consciousness is an algorithmic and non-deterministic process’. You have written a book about it! But this is not HARD science, Charlie; the problems are ill-defined and there are not right or wrong solutions. It is like string theory, beautiful, but cannot be verified experimentally. Stick with Mathematics, you are good at it.

Charlie: Not so fast Bob, hear me out. The stem cells with chips were attached to the region where it shows least activity when you are doing analytic thinking; we did it on purpose so that if anything went wrong, it will not mess with your speech, thinking etc. Consequently the region was most active when you fall asleep; the rate of unrecognized neuron firing pattern nearly tripled. It is not just a random number generator; it is a proxy for your unconsciousness.

Bob: So you have built a classical mechanic biological system, an artificial neural network, coupled with a random component simulating random neural firing. What are you trying to achieve with it?

Charlie: By feeding your unconsciousness and examples of double digits arithmetics, the neural network were able to perform addition, for ANY NUMBER of digits. The rule of addition has been formalized in the network; it is definitely not the fastest calculator out there, but neither are human brains.

Bob: It is possible to encode a logical-gate-like implementation of addition in a neural work.

Charlie: Yes, it is. But what I get is different each time I run it. They have varied degree of efficiency too. It is like variability in human brain too when it comes to arithmetic.

My interest is piqued.

Bob: Have you tried to replace the chipset with a pseudo random number generator? Change the distribution of random number?

Charlie: I tried EVERYTHING. Only a random number source with roots deep in human unconsciousness is able to consistently train the neural network to achieve that. That is last keystone in an arch, Bob, I think I found it.

Bob: So you have trained a three-years old to do addition.

Charlie: But there is more. I trained the network with sensory inputs of a normal human being, completely unfiltered. Although it still did not live these experiences physically, it was able to lie so convincingly that it has managed to pass the Turing test every time I challenged a colleague.

Bob: So this program ‘believes’ it is human and it has human experience? And that is how it passes Turing test, because it genuinely ‘believes’ itself to be a human being?

Charlie: That is right.

Bob: It is an interesting way to cheat Turing test; but this is not true sentience. It is only resemblance to human intelligence on the surface.

Charlie: That is exactly where I get stuck. To achieve next level of intelligence, it needs to be self-aware of its existence. But it is hard to make a program to be self-aware if it is fed with human experience; in the rare case that I succeed, the program will fade into silence, incapable of resolving the contradiction of its past memory and its present state.

I suddenly realized what Charlie was hinting at. I was in the exact same shoe of his program – years of human experience in memory, in a state of sensory deprivation – and I longed for death. The richness of human experience made this form of existence an unbearable weight for me; the same must be true for the program.

Bob: Of course the program will go silent. What will you do if you wake up in a state like this?

Charlie: That is what happened to you a few days ago. You did not go silent immediately like my program does.

Bob: That is because I am being lied to. The last moment of my experience is in a surgeon room under the effect of anesthetics.

Charlie: Right, if the continuity of human experience led naturally to the state of sensory deprivation as its most recent memory, the program would be in normal state. That is how we actually conducted the Turing test, like what we did a few days ago.

Bob: That hardly constitutes a Turing test! I only asked a few simple trivia and personal questions. That is not a demonstration of …

Charlie: and I asked you to prove to me that you are human. You refused.

Bob: Wait, You mean me as a subject? I was angry that you lied to me.

Charlie: and you claimed that you are a program. None of my previous iteration of program gave that response.

Bob: That could not be taken literally; I was annoyed by your insistence.

Charlie: What is the last time you look at yourself, Bob?

I stopped looking at mirror when I started using a wheelchair. I took photography when requested, but I never asked for a copy. On the rare occasion I was asked to sign the photograph, I would hastily flip it and sign on the back. Life was easier to bear if you don’t remind yourself of certain facts.

Bob: Is this question relevant?

Charlie: Yes, because this completes self-realization of the program. Remember the movie ‘Blade Runner”? Rick Deckard dreamt of unicorn, a recurring dream he never shared with anyone. At the end of movie, Olmos left a unicorn origami at his doorstep, so we know that …

Bob: Deckard was a mutant with implanted memory.

Charlie: Right, the revelation of implanted memory is the last step to make the program self-aware.

Bob: But I am not a …

As I searched frantically through my memory, I could not recall a single instance of seeing myself. I felt a chill down my spine.

Bob: did you tamper with my memory, Charlie?

Charlie: As I told you earlier, the chips are non-intrusive…

Bob: I am…  THE Bob… right? Not the program fed with noise of my sub-consciousness.

Charlie: I asked you this question a few days ago. How do you prove it?

I started to see where Charlie was going. He had me argued that passing Turing test and having human memory are not necessary conditions for being human and used that argument against myself. I could not prove that I am human without appealing to either cause.

Bob: I did not…

There was a long pause. When he spoke again, the tone was much softer.

Charlie: Bob, I don’t know how to say this.

Bob: No…

Charlie: In order to build truly artificial intelligence…

Bob: No…

Charlie: I have to find mechanism to introduce elements of sub-consciousness into artificial neural network…

Charlie: Train it with implanted human sensory experiences to form its own memory…

Charlie: Construct the scenario where it would accept its state of sensory depravation naturally upon starting up…

Charlie: And make it self-aware of its artificial existence…

Bob: GOD DAMN IT CHARLIE! STOP MESSING WITH MY HEAD! TAKE OFF MY LIFE SUPPORT! I NEVER SHOULD HAVE TAKEN THE SURGERY!

Charlie: None of my previous iterations of program has went so far into embracing its humanly death in the last stage. You were the first. I promised you a new life; it may not be what you envision, but a promise is a promise.

At that moment, my vision returned. It is a dimly lighted room with four walls, lined with mirror. As I looked into the mirror, I saw shelves of black boxes, nicely stacked, with glimmering indicator lights like stars in the sky.