Back when I was a young whippersnapper in the corporate world, one of my leads asked me if I read a lot of books. Trying to be modest, I said yes, but that I don't read as much as I'd have liked to. This is a fairly common conversation I have with people: someone sees me sitting with a book on my lap and ask me if I read a lot. I say I do and then ask for suggestions. I then ask them what kind of books they've read in the past, and what they think they'll enjoy. Depending on their answer, I give them one of these:
- Don Quixote
- Old Man and The Sea
- Tipping Point
- Guns, Germs and Steel
- Why Nations Fail
- Slaughterhouse Five
- Fahrenheit 451
- A Short History of Nearly Everything
- A Song of Ice and Fire
- American Gods
I think of myself as a recommendation engine with a very small learning dataset. The factors I take into account include things like how much interest they showed in the conversations we've had in the past, the apprehension with which they approached me, the level of specificity they show towards the topic they're okay with and the ones they'd rather avoid, and my understanding of how likely they are to just read the back cover and put it down forever.
The guy who asked me was not just my boss, he was a guy who was absolutely proud of not being the kind to read books. He was outgoing and boisterous, and was generally regarded with healthy disdain. I had only known him for a month or two at this point, but I already knew what I was going to tell him to read: The Monk Who Sold His Ferrari, Thinking Fast and Slow, Drive, Goal! or The Alchemist. I was essentially going to tell him to pull out belly button lint and examine it slowly and pretentiously.
But he didn't give me that opportunity. He asked me a question I've only ever been asked that once: why should I read? I remember buying some time by saying something like "I don't know if you should read, because everybody reads for their own specific reason". The truth was, I had no clever reply. I was blank.
For the longest time, I've read only to read. I've always taken great joy in knowing things. It wasn't always because I wanted to learn something great about the world or because I wanted to do something with all the information in it. I did it just because. I was collecting knowledge for the sake of it. I knew all kinds of "fun facts" like the capitals of Mongolia, Madagascar and South Africa (Ulaanbaatar, Antananarivo and Cape Town, Pretoria and Bloemfontein) some less interesting things like the bird with the longest wingspan (The Giant Albatross), the reason the hominid Lucy is named that (because the people who found it were listening to Lucy In the Sky With Diamonds at the time of excavation and thought that the specimen they found was a woman) to some downright bizarre things like what Green Boots Cave is and why it's named that in the first place. That last one, you'll have to look up to find out.
But the point is, I just read things. I read newspaper snippets my churumuri or bajji came wrapped in, the magazines at beauty parlours, textbooks, dailies in any language I could read - over the course of a summer, I taught myself how to read Telugu just because I wanted to read the newspaper at my grandparents' house - and sometimes just signs and hoardings along the roadside. All this apart from books of every genre and quality I could find.
My dad taught me the value of reading from a very young age. He used to make me read the newspaper aloud so that I didn't stammer or misread anything. He was never loathe to spending on my education and books, even as he would beat me senseless for asking for this or that toy too much. He was a weird man, my dad. But I learnt a lot from him. One of the most valuable was that the only way I could gain his approval was by being better-read and well-learned than he was.
And so, I read. I read to impress him with facts he didn't know, to get into quizzes he would approve of, to outdo him at every level of his life. While I was inspired by him, I also absolutely detested him for the better part of my childhood and teenage years. Like all sons, I considered my dad too strict and overbearing, and thought that he was holding me back from all the fun in the world.
But the full list of my daddy issues is for another post.
The truth is, though, I didn't read for his approval for very long. I quickly outgrew that, and reading just became a thing by itself. I was a kite floating high up in the sky, and if anybody asked me why I was floating, I wouldn't know anything about the kid who ran into the wind for half an hour trying to get me to climb.
But at this particular moment, I told Vimal, in complete honesty, "I read to be able to talk to people. To hold conversations." The answer struck me as much as it impressed him. Not only was this completely devoid of any of that pretentious fluff about growing as a person, discovering the meaning of life, changing people's lives, incorporating great ideas and whatever else I'd spouted before, it was elegant and simple. It wasn't aggrandizing, and what I said made me feel extremely small and petty for doing it for such a silly external reason.
Over time, though, I've some to see it slightly differently. I don't think it's silly or petty that I spend so much time and effort on a conversational tool. That's because I realize now that all ideas and opinions are conversations we have with ourselves, where competing parts of our minds battle over which side of the issue is the right side. What starts as an observation becomes a thought when it passes through our cognitive faculties. Our thoughts are sporadic and unbiased recollections of things as we see them. And they regularly fall on either side of any issue.
When I observe that there was a terrorist attack and the perpetrator was Muslim, that becomes a thought: "Muslim man blew up a whole bunch of people in the name of his religion". When I learn that half of Iraq is up in arms against ISIS, that becomes the thought "Muslim men are fighting each other in several parts of the world". When I read that Muslims have a history of infighting, I think "there's no such thing as a standard-issue Muslim". If I start thinking about whether Muslim men are to be feared, these thoughts may fall on opposite sides of the issue, as will all the individual thoughts I've ever had or picked up from someone else. And given enough time and cognitive room, all the opposing parts of me will go at it, arguing against each other and agreeing with each other and qualifying each other's statements until I arrive at a considered opinion: "Muslim men are really no different from any other religious group". This is how cognition works.
This is the basis of human understanding. This is the reason we have such diverse opinions about the same thing: because we've all arrived at the position with varying degrees of understanding and considerations. We don't all get access to the same information, meaning we all think in slightly different ways, emphasising one aspect of the equation more than the other. We reinforce different parts of our thought processes and penalize them differently. Over time, we no longer hold the same fundamental positions. Someone exposed to people of other ethnicities and cultures from a young age reacts to them differently than someone who learns of them late in life. That's because every thought we hold and every observation we make is tinged with the realizations from previous ones. That's why human thought is unspecialized and fuzzy, but rich in variety. we see thousands of people each day, and interact with hundreds. Everything we pick up along the way, we talk to ourselves about. Constantly.
Which brings me to AI and why it hasn't taken off yet. As much as I would have cherished the trite pun, I mean artificial intelligence; not Air India. We all know why Air India hasn't taken off: the damn pilots are on strike. No, AI's failure to live up to its billing has nothing to do with computational capabilities and government regulations. I'm not discounting how woefully unprepared we are for an AIpocalypse: technologically, socially or industrially.
But the biggest reason why we're still stuck with crummy programs that are largely just line followers on juice is that we haven't been able to figure out a way to simulate the human mind. We've managed to build ANNs that can very well mimic (if not outperform) human brains, but nothing to make up the mind.
The mind is more than just memory and some basic instructions on learning and reward systems. The human mind is as complex as it is elegant. It exists in each one of us in the form of thought, memory, bias, prejudice, the physical brain, the neural apparatus, genomic quirks, copying errors and a "survival instinct" handed down through generations of natural selection.
It exists in all of us, simultaneously, in the form of social rules and constructs, language, faith, fears, values and strong inexplicable instincts that Carl Jung thought constituted our "archetypes".
And yet, it exists nowhere. It does not occupy any space, nor does it make its presence known through something that we cannot directly attribute to the brain. We don't know whence it arrives, but we know what it brings: consciousness.
If we can make AI dream big, talk to each other, interact with one another and most importantly, talk to themselves, we've hit jackpot. If we can essentially simulate what the human mind does with its alternating periods of wakefulness and sleep, we can build systems that not only learn but also build on that learning in creative, vastly more powerful ways. While one period can be dominated by the stimulus circuits and all the various sensors to detect and interact with the world, the other can be to consolidate memory, establish connections with existing memories and use the knowledge for creative thought. This creative thought brings with it a certain amount of chance, since we're essentially handing the mind a blank slate. This chance occurrence, when weighed down and consolidated with the other memories, forms a quasi-memory of lower quality than the rest. We've created a dream. But as dreams get woven into the very fabric of memory itself, the mind will find it difficult to cope up with just memories alone. Thoughts and opinions and ideas will arise as the system tries to build itself up after each dream leads to a slight reconfiguration.
As humans, this is our moment. For the longest time, we've fantasized about controlling an entire world and watching it grow, develop, flounder and fall. This is our opportunity to finally do it. All we need to do is let AI read. And talk to themselves.
From there, our downfall is just a gentle push away.