In Greek myth, the capstone of Hephaestus’s divine laboratory was the female android commissioned by Zeus. To punish humans for accepting the technology of fire stolen by Prometheus, Zeus commanded Hephaestus to fabricate Pandora (chapter 8). Each of the gods endowed the artificial maiden with a human trait: beauty, charm, knowledge of the arts, and a deceitful nature. As the vengeful god’s AI agent, Pandora executed her mission to unseal a jar of disasters to plague humankind forever. She was presented as a wife to Epimetheus, a man known for his impulsive optimism. As we saw, Prometheus warned humankind that Pandora’s jar should never be opened. Are Stephen Hawking, Elon Musk, Bill Gates, and other prescient thinkers the Promethean Titans of our era? They have warned scientists to halt or at least slow the reckless pursuit of AI, because they foresee that once it is set in motion, humans will be unable to control it. “Deep learning” algorithms allow AI computers to extract patterns from vast data, extrapolate to novel situations, and decide on actions with no human guidance. Inevitably AI entities will ask—and answer—questions of their own devising. Computers have already developed altruism and deceit on their own. Will AI become curious to discover hidden knowledge and make decisions by its own logic? Will those decisions be ethical in our human sense? Or will AI’s ethics be something “beyond human?”
Released from Pandora’s jar—much like the computer viruses let loose by a sinister hacker who seeks to make the world more chaotic—misfortune and evil flew out to prey upon humans for as long as the world exists. In simplistic fairy-tale versions of the myth, the last thing to flutter out of Pandora’s box was hope. But in darker versions, the last thing in the jar was “anticipation of misfortune.” And Zeus had programmed Pandora to slam down the lid, trapping foreknowledge inside. Deprived of the ability to anticipate the future, humankind was left with what we call “hope.” As was true of Epimetheus, foresight is not our strong point.
Yet foresight is crucial as human ingenuity, curiosity, and audacity continue to push the frontiers of biological life and death and the melding of human and machine. Our world is, of course, unprecedented in the scale of techno-possibilities. But that unsettling oscillation between techno-nightmares and grand futuristic dreams—that is timeless. The ancient Greeks understood that the quintessential attribute of humankind is always to be tempted to reach “beyond human,” and to neglect to envision consequences. We mirror Epimetheus, who accepted the gift of Pandora and only later realized his error.
In 2016, Ray Crowder, an engineer at Raytheon, created three miniature learning robots. He gave the robots classical names: Zeus, Athena, and Hercules. With neural systems modeled on those of cockroaches and octopuses, the little solar-powered robots were endowed with three gifts: the ability to move, a craving for darkness, and the capacity to recharge in sunlight. The robots quickly learned to travel and soon understood they must venture into excruciating light in order to recharge or die. This seemingly simple learning conflict of these creatures that were made, not born, parallels human “cognitive economy,” in which emotions help the brain allocate resources and strategize. Other AI experiments are teaching computers how human strangers convey goodwill to one another, and how mortals react to negative and positive emotions.3
Since Hawking warned that “AI could spell the end of the human race,” some scientists are proposing that we could teach human values and ethics to robots by having them read stories. “Fables, novels, and other literature,” even a database of Hollywood movie plots, could serve as a kind of “human user manual” for AI computers. One such system is named Scheherazade, in homage to the heroine of The One Thousand and One Nights. Scheherazade was the legendary Persian philosopher-storyteller who had memorized myriad tales from lost civilizations. She saved her own life by reciting these enchanting stories to her murderous captor, the king. The first stories uploaded into the Scheherazade AI were simple narratives that show computers examples of how to behave like good rather than psychotic humans. With the goal of empathetic interactions with human beings and appropriate responses to their emotions, more complex narratives would be added to the computer’s repertoire. The idea is that stories would be valuable when AI entities achieve the human mental tool of “transfer learning,” symbolic reasoning by analogy, to make appropriate decisions without human guidance.4
Computers may be modeled on human brains, but human minds do not work just like computers. We are learning, for example, that our cognitive function, self-reflection, and rational thinking depend on emotions. Stories appeal to emotions, pathos, the root of empathy, sharing feelings. Stories continue to be alive as long as they summon strong, complicated emotions, as long as they still resonate with real dilemmas, and as long as they are good to think with. We have seen how Greeks and other ancient societies told themselves stories to try to understand humankind’s yearning to exceed biological limits and to imagine the consequences of those desires. The insights and wisdom in such myths might deepen our discourse about AI.
Biotechne stories, perpetuated over millennia, are a testament to the persistence of thinking and talking about what it is to be human and what it means to simulate life. We are hardwired to hear, tell, and remember stories. As George Zarkadakis reminds us, stories “are the most powerful means available to our species for sharing values and knowledge across time and space.”5 This raises an intriguing possibility.
Might myths about artificial life in all its forms, like the examples gathered in this book, play a role in teaching AI to better understand humankind’s conflicted yearnings? Perhaps some day AI entities will be able to absorb mortals’ most profound wishes and fears as expressed in mythic musings about artificial life. Perhaps AI beings might somehow grasp the tangled expectations and fears we have of AI creations. Through learning that humans foresaw their existence and contemplated some of the quandaries the machines and their makers might encounter, AI entities might be better able to comprehend—even “empathize” with—the quandaries that they pose for us.
The rise of a Robot–Artificial Intelligence “culture” no longer seems far-fetched. AI’s human inventors and mentors are already building the Robot-AI culture’s logos (logic), ethos (moral values), and pathos (emotions). As humans are enhanced by technology and become more like machines, robots are becoming infused with something like humanity. We are approaching what some call the new dawn of Robo-Humanity.6 When that day comes, what myths and stories will we tell ourselves? The answers will shape how and what our AI creations will learn too.
GLOSSARY
agency. The capacity, condition, or state of acting, operating, or exerting power or energy in a given environment.
android, droid. A mobile robot in human form.
Artificial Intelligence (AI). Intelligence, or mind, displayed by artificial life or machines, analogous to the natural intelligence of animals and humans; capable of perceiving its environment and taking action. AI mimics cognitive functions associated with mind, such as learning and problem solving. “Narrow AI” allows a machine to carry out specific tasks, while “general AI” is a machine with “all-purpose algorithms” to carry out intellectual tasks that humans are capable of, with abilities to reason, plan, “think” abstractly, solve problems, and learn from experience. AI can also be classified by types: Type I machines are reactive, acting on what they have been programmed to perceive at the present, with no memory or ability to learn from past experience (examples include IBM’s Deep Blue chess computer, Google’s AlphaGo, and the ancient bronze robot Talos and the self-moving tripods in the Iliad). Type II AI machines have limited capacity to make memories and can add observations to their preprogrammed representations of the world (examples: self-driving cars, chatbots, and Hephaestus’s automated bellows). Type III, as yet undeveloped, would possess theory of mind and the ability to anticipate others’ expectations or desires (fictional examples: Star Wars’ C-3PO, Hephaestus’s Golden Servants, the Phaeacian ships). Type IV AI of the future would possess theory of mind as well as
self-awareness (fictional examples include Tik-Tok in John Sladek’s 1983 novel and Eva in the 2015 film Ex Machina). Since she is capable of deceit and persuasion, Pandora seems to fall between Types II and III.
artificial life. Systems, beings, or entities that simulate natural life, natural processes; or replicate aspects of biological phenomena; human or animal artifacts brought to life.
automation. The technology by which action is performed without human assistance.
automaton, automata. A self-moving mechanical or constructed device, usually resembling an animal or human, that is not directly operated by an agent. Some automata are machines that perform tasks according to predetermined instructions; some automata can respond with a range of responses to different circumstances.
bionic. Having artificial body parts that amplify human or animal powers.
biotechne. Ancient Greek bio, life, techne, craft, art, science, the application of knowledge to practice.
biotechnology. Technology based on manipulating biological organisms, living systems, or their components to develop, modify, or make products or processes.
black box. Complex entity, machine, or system whose outputs are known or observable but whose inner contents and internal workings are hidden, unknown, opaque, and mysterious to the user.
cyborg, cybernetic organism. A being, usually humanoid, that combines or integrates organic, biological components with artificial technology, a human-machine hybrid, often exceeding human capabilities.
device. An object, gadget, instrument, contrivance, or apparatus made for a particular purpose, often denoting a mechanical item.
fembot. A robot in the form of a human female.
machine. A mechanical structure or device based on one or more components (such as lever, pulley, wheel and axle, inclined plane, screw, wedge) that changes the direction or magnitude of a force.
machine learning. Computers and AI with the ability to learn independently, without being explicitly programmed.
mechanism, mechanical. Something made of parts that move or work together to perform an action; a machine or something resembling a machine.
programmed. Supplied with a predetermined set of (coded) instructions for automatic performance.
puppet, marionette, doll. An artificial model of a human or animal typically moved by hand, rods, wires, or strings.
rejuvenation. To make a living being young again, to restore youthful strength, vigor, and/or appearance.
robot, bot. Complex and ambiguous to define, but a robot usually is a machine or self-moving object with a power source that provides energy. It can be “programmed” to “sense” its surroundings, and has a kind of “intelligence” or way of processing data to “decide” to interact with the environment to perform actions or tasks. Talos, the bronze animated statue powered by ichor, fits this definition.
Uncanny Valley. The eerie and repellent sensation experienced by most human beings when encountering artificial life forms, especially humanoid entities, that appear to be almost but not precisely real. Affinity increases with verisimilitude but steeply drops off as the entity approaches being indistinguishable from reality. The hypothesis was first identified by roboticist Masahiro Mori in 1970.
NOTES
CHAPTER 1. THE ROBOT AND THE WITCH: TALOS AND MEDEA
1. Apollonius Argonautica 4.1635–88; Apollonius (Hunter trans.) 2015, 6, 298–304. The Greek word automaton, “acting of one’s own will,” was first used in Homer Iliad 5.749 and 18.371-80 to describe the automatic door opening and automatic wheeled tripods built by Hephaestus for the gods; see chapter 7. Hound and javelin, Ovid Metamorphoses 7.661-862.
2. On the “slippery” terms robot and automaton for an ancient “object constructed to move on its own,” see glossary; cf. Bosak-Schroder (2016, 123, 130–31), who argues that the earliest automata in Greek literature were originally imagined as solely magical and only later attained mechanical life. The idea of automated tools that can finish a task without continued human input, along with the impulse to make them, is very ancient, beginning with the Stone Age atlatl (spear thrower) and the bow and arrow. Once the arrow is nocked, aimed, and released, the bow fires “this little spear further, straighter, and more consistently than human muscles ever could,” remarks Martinho-Truswell (2018).
3. For a classicist’s perspectives on Harryhausen’s Talos: Winkler 2007, 462–63.
4. Hesiod Works and Days 143–60. In Hesiod’s poem, the “Age of Bronze” was a symbolic chronology of the warlike Bronze Age generations that preceded present-day Iron Age humans; Apollonius’s poetic license makes the men of that age literally of bronze. Gantz 1993, 1:153. There was also a legendary Athenian inventor named Talos; see chapter 5. Various genealogies of Talos: Buxton 2013, 77–79.
5. Ancient Colchis is now the Republic of Georgia. “Medea’s oil,” Suda s.v. Medea.
6. Apollodorus Library 1.9.26; Apollonius Argonautica 3.400–1339.
7. Medea’s technai, devices: Pindar Pythian 4.
8. Another version of Medea and her relationships with Jason and the Argonauts: Diodorus Siculus 4.45–48. Motif of heroes’ and monsters’ sole vulnerability, Buxton 2013, 88–94.
9. Colossus of Rhodes, Pliny 34.18; Strabo 14.2.5. N. F. Rieger in Ceccerelli 2004, 69–86. Centuries earlier, Rhodes was also famous for its “living statues”; see chapters 5 and 9.
10. Why people tend to attribute life to machines and Artificial Intelligence, Bryson and Kime 2011; Shtulman 2017, 138; Zarkadakis 2015, 19–23, 25–27. Trust and empathy in human-robot interactions: Darling, Nandy, and Breazeal 2015; Lin, Abney, and Bekey 2014, 25–26; and Lin, Jenkins, and Abney 2017, chapters 7–12. When “thinking machines express anxiety about their own demises” it is “surely a sign of ‘consciousness’”; Mendelsohn 2015. Can Artificial Intelligence be tricked? Reynolds 2017.
11. Sophocles Daedalus fr. 160, 161 R. Winkler 2007, 463.
12. In a story mentioned by Apollodorus (Library 1.9.26), the Argonaut Poeas shot Talos in the ankle, which recalls the death of the mythic hero Achilles by a poison arrow to his vulnerable heel. Rock-throwing giants were a common motif in ancient myth and art. Another source says Talos was a bronze bull, perhaps conflating him with the Minotaur, the bull-headed man kept by Minos in the Cretan Labyrinth (see chapter 4). Coins of Knossos show the Minotaur throwing stones, and some Talos coins of Phaistos show a bull on the reverse.
13. Ganz 1993, 1:365. Robertson 1977. Teardrop: Buxton 2013, 82 and fig. 3 caption. Metallic objects and statues were often painted whitish in red-figure vase iconography; for example, several images of Niobe being turned into stone show her body partly white. Another notable detail is the ornamental border around the top of the Ruvo krater that appears to represent blacksmith’s tongs; see figs. 7.4 and 7.5, and the similar design in the border at the top of the Niobe Painter’s krater depicting Pandora, who was also fabricated by Hephaestus, fig. 8.7.
14. Robertson 1977, 158–59. Buxton 2013, 81 and figs. 4–6.
15. Carpino 2003, 35–41, 87, quote 41. Medea and local Etruscan versions of Greek myths, de Grummond 2006, 4–5.
16. Gantz 1993, 1:341–65, on artistic and literary sources for Talos; Apollonius Argonautica 4.1638–88; Simonides fr. 568 PMG; Apollodorus Library 1.9.26 and J. Frazer’s note 1; 1.140; Photius Bibliotheca ed. Bekker, p. 443b, lines 22–25; Zenobius Cent. v. 85; Eustathius scholiast on Odyssey 20.302. Divine robotic devices are discussed in chapter 7.
17. Faraone 1992, 41. Quotes, Hallager 1985, 14, 16–21, 22–25. Cline 2010, 325, 523. For photos and a drawing of the Master Impression seal, Chania Museum of Archaeology, Crete, see CMS VS1A 142 at Arachne.uni-koeln.de.
18. Shapiro 1994, 94–98, on the lost Argonautica epic cycle.
19. Simonides fr. 204 PMG; scholion to Plato Rep. 337a. Blakely 2006, 223. Sardinia and Crete, Morris 1992, 203. Etruscans and Nuragic Sardinian links: http://www.ansamed.info/ansamed/en/news/sections/culture/2018/01/08/etruscan-settlement-found-in-sardinia-for-fi
rst-time_288c45c9-9ae3-4b5e-ab8d-cb9bf654b775.html.
20. Laestrygonians are also described by Apollodorus Epitome 7.13; Thucydides 6.2.1; Hyginus Fabulae 125; Ovid Metamorphoses 14.233; Strabo 1.2.9. A pair of wall paintings, ca. 50–40 BC (Vatican Museum, Rome), depicts the Laestrygonians as copper-colored giants wresting up boulders and heaving them at Odysseus’s sailors. Paratico 2014.
21. Kang 2011, 15–16, 19, 21, 312nn1–3.
22. Weinryb 2016, 154.
23. Gods don’t use technology; Talos is “biological” and not an automaton because an automaton must have “an internal mechanism,” Berryman 2003, 352–53; Aristotle on automaton “self-moving” puppets, 358. Devices made by Hephaestus are “animated by divine power,” not technology, and gods do not use technology, Berryman 2009, 25–26 (Talos is omitted from discussion). Cf. Kang 2011, 6–7 and 311n7. But see De Groot 2008 and Morris 1992 on the overwhelming evidence from ancient literature and art that Greek gods were imagined as using technology and tools in projects, including self-moving entities. “Mechanistic” analogies could arise before “full-fledged automata” were feasible.
24. Bosak-Schroeder 2016, 123, 132. Cf. Berryman 2009, 22, “mechanistic conceptions” could not have been imagined before mechanics developed “as a discipline.” Contrast Martinho-Truswell 2018 on prehistoric inventions and see Francis 2009; archery, catapults, voting machines, and the winepress demonstrate practical mechanics.
Gods and Robots Page 23
