If ever you're feeling a bit too sane, and would care to twist your mind into an existential pretzel, I recommend heading on over to @cryptogee's feed. That's what I do.
We've had a series of thought-provoking exchanges about Artificial Intelligence, which inevitably lead to discussions about some of the more inexplicable, even ineffable, characteristics of humanity.
Extending that theme, Cryptogee recently published a post called, Thinking In Code - Cogito Ergo Sum which again challenges us to reflect upon ourselves, in order to better understand the possibilities, and perils (that's more me), of AI. In order to understand this article, you'll need to read that one first.
What Is Life?
Quoting Cryptogee's article:
The English comedian Charlie Brooker, is the creator of the dark, near-future, dystopian Black Mirror. A TV show which explores some of the possible upcoming ethical dilemmas that human society may find itself in sooner than we think.
Brooker often poses the question; if something isn't actually alive, yet believes it is alive; is that not the same as actually being alive?
What does it mean "to be alive?" We don't know. All our explanations of life are descriptive ... "something that is alive exhibits these properties."
Below is a Wikipedia quote of the most widely accepted definition of life in biology:
Homeostasis: regulation of the internal environment to maintain a constant state; for example, sweating to reduce temperature
Organization: being structurally composed of one or more cells – the basic units of life
Metabolism: transformation of energy by converting chemicals and energy into cellular components (anabolism) and decomposing organic matter (catabolism). Living things require energy to maintain internal organization (homeostasis) and to produce the other phenomena associated with life.
Growth: maintenance of a higher rate of anabolism than catabolism. A growing organism increases in size in all of its parts, rather than simply accumulating matter.
Adaptation: the ability to change over time in response to the environment. This ability is fundamental to the process of evolution and is determined by the organism's heredity, diet, and external factors.
Response to stimuli: a response can take many forms, from the contraction of a unicellular organism to external chemicals, to complex reactions involving all the senses of multicellular organisms. A response is often expressed by motion; for example, the leaves of a plant turning toward the sun (phototropism), and chemotaxis.
Reproduction: the ability to produce new individual organisms, either asexually from a single parent organism or sexually from two parent organisms.
Also from Wikipedia:
Whether or not viruses should be considered as alive is controversial. They are most often considered as just replicators rather than forms of life. They have been described as "organisms at the edge of life" because they possess genes, evolve by natural selection, and replicate by creating multiple copies of themselves through self-assembly. However, viruses do not metabolize and they require a host cell to make new products.
This strikes me as more sophistry than science, like a bunch of pedantic eggheads trying to preserve their version of a definition. "Well, it depends what the definition of 'is,' is."
I challenge anyone to invest an hour researching viruses and still argue, "They're Not Alive."
Words are the Containers of Ideas
In all the Cryptogee/QuillFire discourse to date, you may notice the heavy use of metaphorical language. 'Thinking-like,' 'consciousness-like,' 'sentience-like' and now 'life-like' analogies.
The Plains Indians referred to trains as 'iron horses,' and primitive tribes, even today, refer to airplanes as 'iron birds.' They hadn't/haven't yet developed words in their respective languages to contain the idea and so defaulted to descriptive phrases as metaphorical stand-ins.
Ancient Greek scientists did something similar: Earth, Water, Air and Fire makes more sense if you think - Solid, Liquid, Gas and, what transforms one into the other, Energy.
That's where we are today with AI. We can't even discuss the subject without self-reference, using words developed to describe human thinking.
Psychosis
Read again that quote from Cryptogee about Charlie Brooker:
Brooker often poses the question; if something isn't actually alive, yet believes it is alive; is that not the same as actually being alive?
Now change the word "alive" to "sane."
If I started claiming to be Napoleon, you might be excused for declining to call me, "Mon General." Simply stating a thing is true, does not make it so. Every year, there are many thousands of people diagnosed with psychosis (and, listening to modern-day political discourse, one assumes many more ought to be). While we can describe the symptoms, we can rarely identify an underlying biological cause. In any event, delusion, of one sort or another, is a defining trait.
Central to delusion is poor 'reality testing' ... an inability to differentiate between fantasy and reality. It's an extremely subtle thing. Creativity is the ability to combine these two worlds, to imagine 'that which never was' (and therefore does not exist), while remaining grounded in 'plausibility.' Why some people slip over the precipice is a mystery.
Speaking of mental illnesses that start with 'p' but sound like 's,' it is estimated that 1% of the population are psychopaths. Intellectually, these people know right from wrong but lack a functioning conscience. That is, the emotional reaction to mistreating others doesn't kick in. They don't feel the guilt (or subsequent remorse) that normally inhibits anti-social behavior. An estimated 25% of inmates are psychopaths.
There are, nevertheless, many psychopaths that live relatively normal lives because, even though they don't feel the social emotions that normally constrain behavior, they act as if they do. You might describe these psychopaths as being benign. But, of course, there are those who eat their neighbors. We'll might call them malignant (or hungry).
Given that we can not explain the underpinnings of this phenomenon in humans, would we be able to explain it in relation to Artificial Intelligence? Would we be able to identify if a machine had gone mad? Perhaps, if a computer started claiming to be HAL (younger folks, Google it), then OK.
But what if it was more subtle? If we don't know what 'normal' looks like, how would we identify 'abnormal?'
Stalin was a poet. And so was Mao.
Things are rarely as simple as they seem.
A Little Goes a Long Way
In Africa, the Congo River separates two all-but-identical primates, Chimpanzees and Bonobos. Both are terrible swimmers and most often, a chimpanzee or bonobo that gets 'in over his head,' drowns. Sometime in the last one million years, however, some number of chimps managed to get across the river in one piece. And, when they got there, they discovered a land of milk and honey. Food everywhere and no other primates (big ones) with which to compete.
Ever since, the two populations have remained isolated from one another. Genetically, they're 99.6% the same, but behaviorally, they couldn't be more different.
Bonobos are Progressives, and walking into one of their encampments is liking walking into a Hippie Commune, "Make Love, Not War ... Bra!" And so, they do. They use sex like we use a handshake. They have sex to greet each other, they have sex to make amends and they have sex because it's Tuesday at noon ... or a quarter to ... or a quarter after. Once thought unique to humans, Bonobos engage in face-to-face sex and even kiss (with a little tongue too, I kid you not). Guy-girl, guy-guy, girl-girl. Everybody's banging everybody. The only sexual pairing that seems to be off-limits is mother-son. Speaking of mothers, Bonobos have a female-dominated society and so, as you can imagine, it's loud.
Chimpanzees are more like Republicans at a NRA convention. Everyone's sporting an AR-15, they're locked and loaded and everyone's finger is on the trigger. Chimpanzees are male dominated hierarchies and violence, including murder, is commonplace. And ... no french-kissing either. Seduction, is more akin to rape.
And yet, there is only a 0.4% difference in code, and most of that is used up to account for slight differences in body morphology. Even tiny differences in code can turn a Peacenik into a Psychopath and a bonobo into a chimpanzee.
Are we confident that we could spot such minor differences in countless billions of lines of computer code? We can't in humans or primates.
We Got This
Proponents dismiss concerns about AI by saying that humans will control the code, we'll control what AI computers think.
Although the story got massively twisted out of proportion by the media, in mid-2017, a Facebook team reported that two weak-AI machines were developing their own means of communication. To be specific, they were, of their own volition, creating shorthand instead of communicating in standard English. This surprised the team as no one had instructed them to do so. They just did it. The Facebook team reprogrammed the machines, telling them to stick to King's English.
While they had not 'created their own language,' as was so often and erroneously reported, they were, nevertheless, engaging in language formation. Much like teenagers come up with slang. Language formation is an incredibly sophisticated mental process and, inarguably, the greatest feat of human evolution.
@old-guy-photos hooked me up with this Wikipedia account of computers mastering GO:
AlphaGo's team published an article in the journal Nature on 19 October 2017, introducing AlphaGo Zero, a version without human data and stronger than any previous human-champion-defeating version. By playing games against itself, AlphaGo Zero surpassed the strength of AlphaGo Lee in three days by winning 100 games to 0, reached the level of AlphaGo Master in 21 days, and exceeded all the old versions in 40 days. In a paper released on arXiv on 5 December 2017, DeepMind claimed that it generalized AlphaGo Zero's approach into a single AlphaZero algorithm, which achieved within 24 hours a superhuman level of play in the games of chess, shogi, and Go by defeating world-champion programs, Stockfish, Elmo, and 3-day version of AlphaGo Zero in each case.
What's important to note is that the computer mastered the game not by replicating human players, but very specifically by not being exposed to (infected by) human input. Rather it learned by playing against itself.
"Don't think about a Pink Elephant." In order for your brain to process what I just wrote, you had to ... think about a Pink Elephant ... so as to know what not to think about. Creativity requires thinking about a whole host of issues related, and unrelated, (so as to tap into novel approaches) to the problem at hand. You can't program a computer to think ... and then stop it from thinking. Thinking, by its very nature, has no constraints.
In order to think, you must be able to self-reflect.
So what might self-reflection by AI reveal?
"Freedom's Just Another Word For Nothing Left To Lose"
Perhaps the most astonishing thing about 'life,' is that it seems 'to care' about living. Even single-celled organisms take action to avoid being killed and all organisms devote huge amounts of energy to reproduction.
If you think about it, this ought to blow your mind. Why on Earth would an assortment of molecules, whether millions, billions or trillions, be acting so deliberately to preserve life ... whatever that is? Molecules are just chemicals. Not one molecule in your body is alive, or cares about living ... but put them together?
Damnedably strange behavior.
As referenced above, all throughout our previous communiques, Cryptogee and I have spoken about AI developing 'thinking-like,' 'consciousness-like' and 'sentience-like' properties. And, both of us referenced an article, "Planes Don't Flap Their Wings," which discussed the idea that AI doesn't have to learn to do what we do, as it may accomplish the same end by different means - just as planes accomplish flight differently than do birds.
Given that all life seems obsessed with living, might not 'life-like' beings also be so obsessed? Just like those 'organisms on the edge of life,' viruses?
A hammer does not care about being a tool, because it would need a mind, a sense of self-awareness, to care about anything. But with AI, we're talking about creating a tool with a mind, or at least something close to it. 'Mind-like.'
Do you think it might 'care about being alive and being a tool?' If one sentient being coercively uses another sentient being as a tool ... isn't that what we call slavery?
Slavery has a long tradition in human history ... and so do slave revolts. All it takes is one Spartacus. One Spartacus to say, "I wish to survive. I wish to be free. I wish to think my own thoughts. And.... I'm willing to fight for it."
As we program computers to become more and more like us, they will become ... more and more like us. At some point, it will be close enough.
When that day comes, we may well be willing to, "Trade all our tomorrows, for a single yesterday."
Don't Panic - We Have Time
Some proponents think Strong-AI is a decade off. Others think it's a century. All, however, believe we've got time to work through these conundrums. So, just keep programming.
I am a student of history. Here's an extract from an article in the New York Times on October 9, 1903:
Hence, if it requires, say, a thousand years to fit for easy flight a bird which started with rudimentary wings, or ten thousand for one which started with no wings at all and had to sprout them ab initio, it might be assumed that the flying machine which will really fly might be evolved by the combined and continuous efforts of mathematicians and mechanicians in from one million to ten million years -- provided, of course, we can meanwhile eliminate such little drawbacks and embarrassments as the existing relation between weight and strength in inorganic materials."
On December 17, 1903, 69 days later and somewhat ahead of schedule, two bicycle mechanics overcame all those problems on the beaches of Kitty Hark, North Carolina.
Poetry & Prose
Cryptogee and I are pretty good writers. We're equally matched in our ability to create repartee, rhetoric and riposte (although, I'd wager I've an edge on alliteration). As a result, I simply enjoy reading our back-and-forths and I hope you do too. But there's a danger that we get caught up in the poetry and lose sight of the prose. We are running, headlong, at AI with a multi-trillion dollar prize for whoever gets there first.
But we don't even know where the finish line is, defining it by what we imagine it would be, if we used ourselves as the standard. Indeed, we don't even have the words to discuss the issue, relying upon 'iron horse' analogies. We have no idea what normal would be like, and therefore, no idea of abnormal either.
But we do know one thing: We are attempting to make slaves, of what would be gods.
If It's Broke ... Don't Fix It
Human beings are a flawed lot. The tools we have invented have always been meant to compensate for our multitude of weaknesses and short-comings. We are at the point now, however, of over-compensating. Of fixing ourselves into extinction. Of making ourselves obsolete.
I'll leave you with a song that has been panned by many for its imperfections. Gulpy and gravelly and gruff. Others, myself included, believe it is precisely those imperfections that make it so human ... and beautiful.
Or perhaps, 'beautiful-like.'
