Spiritūs in Machinis
Staring at infinity when it stares back
The philosopher Liezi, in texts dating from the 4th century BC, tells an apocryphal tale of an engineer named Yan Shi, who presented a fully articulated humanoid automaton to King Mu of Zhou (~10th century BCE). It could move its limbs, sing, and even flirt. The king was both impressed and freaked out—so much so that he ordered it taken apart.
In the Western world, the Greek engineer Hero of Alexandria (1st century AD) described mechanical birds and automata powered by air pressure and water. During the Islamic Golden Age, inventors like Al-Jazari built elaborate water clocks and mechanical birds. In the 18th century, Europeans built wind-up automata of birds that could chirp, flap wings, and turn heads. Think Fabergé-level complexity.
Swiss watchmaker Pierre Jaquet-Droz built mechanical dolls capable of writing in cursive with ink pens, playing music on tiny organs, and drawing pictures with quills. These weren't toys. They were programmable, gear-driven machines—18th-century robots with hand-cut cams. Some still work today.
Humans have been creating automata and clockworks for centuries, even millennia, that mimic organic functions, in scales from miniature to bigger than life. A number of companies are threatening to release androids to the mass market within the next couple of years. We may soon be living in an Azimovian dystopia of life-like machines doing our drudge work and hazardous duties.
As an aside, the word “android” is composed of Greek ἀνδρ- (andr-) = man, male human, and Latin -oid = like, resembling (from Greek -oeidēs, "form or likeness"). The term was coined around 1728 by Albertus Magnus in reference to artificial humans or automaton. One wonders if we will also see “gynoids” (nod to Gwyneth Jones) devices, too?
We should also note the meaning of the word “sentient,” a word often misunderstood and misused. It comes from the Latin sentire, which simply means “to feel,” as in perceiving things with the senses. It is related to “sentiment,” which is “a feeling”. Note that it has nothing to do with “thinking”.
The latest iteration of these clockworks is the “artificial intelligence,” or AI. AI is essentially a glorified search engine that functions with natural language queries. It is able to assimilate vast amounts of data extremely fast, and present it in a digested form according to the style and format specified. It can analyze natural language using grammatical and stylistic rules, even determining the correct definition of a word like “can,” with multiple unrelated meanings, from the context and syntax of a sentence.
AI is not sentient, nor intelligent, nor even self-aware and conscious. It can emulate certain emotional responses by filtering through subtext, jargon, slang, and other databases, but it doesn’t “try” to elicit an emotional response, it simply copies the most common responses found in its databases scraped from the vast amounts of information on any given network.
As an example, I run my own AI on my system, called FreedomGPT. I have been experimenting with creating unique language models. One of them has been trained on the cumulative works of Shakespeare, and can respond to any query in a style and vocabulary that is Shakespearian, down to the iambic pentameter. I’m also working on a model trained with Zen haiku.
In both cases, some of the responses are very interesting, even spooky, but the vocabulary is limited to that used by the source material, and for the most part the responses are derivatives rather than new creations. The language models do not “feel” the emotions they elicit, they simply rearrange existing information, apply a given set of rules, and pump out variations.
And they do it remarkably fast, though the AI frequently maxes out all 6 of my Ryzen 7500 processor cores and uses 100% of the 32GB of DDR5 available.
At no point is AI, an android, or any other form of automata “alive” or “conscious”. They do not feel anything. They cannot create anything. If you leave them alone in a room, they will not entertain themselves or play with imaginary friends of their own accord—though you can command them to emulate the behavior within a set of parameters.
The operative term here is “conscious”. Machines are incapable of being conscious. In fact, thousands of years of philosophy and drug-induced ramblings have yet to adequately define it. If we can’t define what conscious is, then there is no way to imbue a machine with that quality.
The machines only understand a set of instructions and execute them in a particular sequence. Unless someone has figured out how to create consciousness using a specific set of instructions, then a machine will never have this quality. We might perceive a certain response as being “conscious,” but we are simply projecting an interpretation onto the response. The machine didn’t intend it; it is simply the result of executing instructions. Accidentally on purpose, we might say.
I believe, having pondered this question at some length, that the Universe itself is conscious. By virtue of some combination of brain, spinal column and DNA, living creatures are able to “channel” some aspect of that Supreme Consciousness, though none of us is able to access all of it. In fact, our heads would probably explode if we did.
A machine can be affected by consciousness. Experiments like the Princeton EGGs have shown significant effects on randomness by nothing more than life existing. The field of noetic science can only guess at how consciousness arises, much less find a way to imbue non-living objects with the quality.
Suffice to say, we are in little danger of machines taking over the planet. They may act autonomously, even interact in real time with the environment, but they are not alive and conscious in any way we can define it. They cannot initiate an action of their own accord. They are not self-motivated. They can’t create anything, nor can they have insight, intuition, or sudden revelations.
If you start a session with an AI, then leave it for a month (or year, or century), it will not spend the intervening time thinking of new responses to your input. It will simply sit there until you interact with it again, and it will pick up exactly where it left off. It will never say to you, “Hey, I thought of a new/better way to answer that.” Every answer is its best answer, given the parameters it is working with.
Will AI take over the world? Not in this Universe. It has no initiative or ambition. That’s not to say a human won’t use it as a tool to aid in his/her ambition to rule the world, but the AI itself has no such thoughts.
In every conceivable case, the AI is little more than a fancy (and expensive) screwdriver. The screwdriver will never go in search of a screw to tighten, but when wielded by a human it has some very useful applications.
It is conceivable that competing human interests, each with their own AI toolbox, might go to war with their tools, but it is not the tools themselves that initiate the conflict, it is human ambitions and instructions that cause the tools to act.
Where all of this pondering gets really interesting is when we ask the question, “Are we AI tools for some higher power?”
Do we humans really have “free will,” or are we simply following incredibly sophisticated sets of instructions, sometimes with unpredictable outcomes based on randomness, but always at the behest of a Great User in the Sky?
And here we come back to the issue of consciousness. Our best minds throughout our recorded history have tried to answer what it is, without satisfactory answers…yet. Until we know what consciousness is, we will never be able to give that quality to our toys.
Is it possible that we cannot answer the question of what consciousness is, because we are not in fact conscious ourselves?
Have fun chewing on that one.
=====
Today’s cultural feast is Solaris (1972)—no, not the George Clooney rubbish. This Polish film does the best job possible to capture Stanislav Lem’s 1961 novel, and anything by Lem should be on everyone’s bookshelf. Both the novel and the film explore the human spirit trampled under Socialist boots. Prepare yourself for a 3-hour deep dive on the nature of existence and consciousness. Bonus culture: listen to Jean-Luc Ponty’s Enigmatic Ocean while reading the novel.
Picking naval lint on the Far Side:
E-book: Paper Golem: Corporate Personhood & the Legal Fiction
Contact Bernard Grover: bernard (at) radiofarside (dot) com
Radio Far Side, published (mostly) every Sun/Wed at 7a CST/7p WIB, is a labour of love. We don’t use a paywall, and we don’t sell stuff. We just create things to inform and entertain. But like any good busker on the digital mean streets, we put our hat down and if you feel inspired, drop a coin in to show your appreciation:
BTC wallet: bc1qth6drgzcyt7vlxxpvqh6erjm0lmaemwsvf0272
XRP wallet: rMSQzLyE3RHacCLwYPADBbq4RHQ71HpCzw
I agree with your view on AI as a useful tool and one that will never achieve true consciousness. However, it is possible that a malevolent actor could program an AI to seek out and destroy bank accounts or digital information of their enemies in order wreak financial havoc. As you wrote, you point an AI in a direction and it just goes until it achieves the programmed end. I don’t fear Skynet evolving, I fear rouge actors using AI to destroy things.
As for consciousness, I pay taxes. I exist. The government sure notices when I don't pay.