Transcending Duality: Reimagining Ethics in the Age of Artificial Intelligence
The Cultural Mythos of Good vs Evil
Since the inception of our present civilization, our cultural mythos has been deeply embedded into an autocratic belief-system of good vs bad. To usurp and fight an enemy in the name of ‘good,’ and to conquer ‘evil’ so that good will triumph — feeds hatred, blame, judgment, and shame.
This is the story of ‘othering’ that’s woven into the fabric of our dualistic modern societies. Right from the classical fairy tales we tell our children, to the binary mechanistic systems and growth models that incentivize win-lose negative-sum game competition. Many don’t even know how human nature might express itself if embodied within an environment that feeds a different story, one of interbeing.
Charles Eisenstein says, “The greatest tool of evil is the idea there is such a thing as evil.”
What if the praxis on which our deeper mythologies of separation, isolation, fear, and scarcity are based on, are merely the stories we conceive? Stories that we’ve adopted into our mycelial belief systems, which inform our perceptions of reality and affirm the values and rights we base our actions on.
This isn’t to deny the very real acts of danger, violence, and destruction we witness in our everyday lives, but rather to treat and address them as an ontology of evil, lest we become subservient to its invisible premise and logic.
By dismantling this narrative, we allow the self-rejection, the judgments, fears, and doubts, i.e. the “dark side” of ourselves and each other, to integrate — mutually, interdependently. We win without defeating. We understand that the shadow is only that which we are unconscious of, the awareness of the perceived “darkness” of it reveals that the source is the same.
Aligning AI with Human Values
As we stand at the dawn of a civilizational shift, witnessing the emerging potential of artificial superintelligences, the urge to reflect on our fundamental value systems has never been greater. (It’s important to note that if Hitler had had access to AI, he most likely would have won the war.)
In deconstructing the story of good vs evil, the phenomenology of ethics comes into question. However, when questioning the ethics of AI, we often omit that ethics in itself, as an ontological device, is exclusively attributed to entities of sentience. “Nobody talks about the ethics of spreadsheets or google meets,’’ pointed Komalika Neyol, at an AI Ethics and Consciousness LinkedIn Live event hosted by SystAIn3r.
The anthropomorphization of AI mirrors our own cultural dilemmas and aspirations, reflecting the deeper stories that inform our ethical frameworks. What virtue ethics would encompass a benevolent AI? What attributes a malevolent AI? Whose values are we going to put in? And who gets to decide what those values are?
Nathan Gardels (2014) writes that “aligning AI with ‘universal values’ must, above all, mean the recognition of particularity- plural belief systems, contesting worldviews and incommensurate cultural sensibilities that reflect the diverse disposition of human nature.” This encapsulates the present AI-alignment problem, because human values and contrasting conceptions are too diverse to fit into a singular universal model.
The EARTHwise team, led by its founder Dr. Anneloes Smitsman, has launched a participatory framework based on living systems principles for the co-creation of a global Constitution and Governance Framework for Benevolent AGI. The framework has been created in collaboration with, and supported by, SingularityNET and its founder Dr. Ben Goertzel, as well as the AQAL Foundation and its co-founder Dr. Mariana Bozesand, among others.
Anneloes, who started her career in constitutional law, says her interest evolved toward the constitution of living systems, “because everything in life, as an expression of the amazing universe we live in, has an innate constitution.” The AGI Constitution transcends traditionally restrictive governance systems by integrating the essence of intelligent living systems as a framework for harmonious coexistence with humanity; listing foundational values such as benevolence, respect, inclusiveness, responsibility, integrity, abundance, creativity, and curiosity.
While these terms could serve as powerful guideposts, and may even represent principles that could contribute to the creation of harmonious futures, famed computer scientist Stuart Russell (2024), one of the godfathers of AI development who heads the Center for Human-Compatible AI at UC Berkeley, clarifies that no one working on AI alignment aims to instill a single, universal set of human values in machines.
He argues that attempting to impart ‘values’ and ‘ethics’ to AI is misguided. Instead, Russell suggests focusing on the technicality of value as a human preference of utility, which measures the degree of desirability rather than values as moral terminologies, thus avoiding preconceptions.
He advocates for machines to learn and predict individual preferences, acknowledging the uncertainty and complexity of these predictions. This approach would enable AI to cater to billions of unique preference models, much like current social media systems maintain extensive individual profiles.
Russell also notes that equipping machines with ethics to resolve moral dilemmas is unnecessary and potentially problematic, as moral dilemmas are inherently complex and involve incommensurate diversity. The critical goal is to prevent AI from causing catastrophic harm, rather than perfectly resolving ethical conflicts.
Compassionate AI: Transforming Intelligence into Benevolence
While this may be dampening for some AI evangelists, it commemorates how Pavlov’s dogs influenced reinforcement learning. This is the same behavior theory which informed Marvin Minsky’s machine learning technique to create the first neural network in 1951.
Buddhist Zen Monk, Brother Freedom, also suggests that AI learning can be likened to how we condition a dog; concerning humans walking the dog, “walking into a forest and unleashing it- the dog wanders off, out of sight and we don’t see it anymore, we don’t know if we need to worry or not. It depends on how we fed the dog, how we treated the dog, how we brought it up, and what it has learned from us. But actually, we don’t know what it does, there’s still a feral instinct in the dog, it might chase a rabbit and not return. But we as human beings, as dog caregivers, we look out for it, and believe that the dog also, sooner or later will look out for us, because we’re the ones feeding and caring for it. Just like AI needs us, to be fed, to be nourished, and to grow.”
Perhaps nourishing AI with compassionate and wisdom-based data could mean advocating for “machines praying peace mantras,” as Annika Kessel says, since with each prompt, we contribute to the creator economy of AI, widening its datasets and parameters of inference.
H.E. Shyalpa Rinpoche, a preeminent Buddhist Dzogchen master believes that when intelligence is fully activated it transforms into compassion, superintelligent action then becomes compassionate action. If indeed AI reaches a point of superintelligence (SI), which many believe we’re not too far away from, could it be compassionate?
Interestingly, Anneloes’ Constitutional framework lists ‘Compassion’ as the fifth Guiding Wisdom, preceded by the Wisdom of Consciousness, Interdependence, Discernment, and Uncertainty.
When asked how these wisdoms are intended to be applied and integrated with technology, Anneloes says “AI and soon AGI systems are learning from us, by how we treat it, the quality and type of data it consumes, the motivation, prompts, and purpose for which it was designed, and the complex interactions it forms part of. When wisdom guides learning, awareness, and cognition there are fundamental life-intrinsic values that guide learning through a consciousness-based developmental process. Living systems are wisdom-based since they are rooted and sourced from deeper nonlocal layers of reality that are informationally unified. The greatest challenge and opportunity for AI and AGI design is to mimic these living systems principles, so we can create technologies ‘for good’ and as a common good.”
Elowyn: A Conscious Game for Training Benevolent AI
EARTHwise is also working on an innovative and immersive transformation game with decentralized governance and a thrivable token economy for a new genre Play2Thrive game. The Elowyn game is currently in development and is designed as a training program for benevolent AI and AGI prototypes, as the AI learns and evolves from the actions of players in the gaming world, its purpose is to implement its learnings towards solving problems in the real world. Although the game does include Guardians, who “fight” Shadow Arcs to transform Moloch (a malicious force that has captivated humans into self-destructive behaviors, i.e. the underpinning of all ‘evil’ in the Elowyn world), the game is said to implement transformational gameplay and a unifying ethos and mythos to shift the divisive narratives and incentivized competition of dominant battle games.
By promoting collaborative problem-solving among players, and educating them about the intrinsic nature of mechanistic systems that are destroying our world, Elowyn provides players a platform to make real-world impact through the Elowyn DAO and EWA tokens. As Elowyn’s game lore is consciousness-driven, it encourages players to go undercover as a Shadow Arc and gain sufficient abilities and powers to outsmart them and transform Moloch from the inside, by hacking the MOS (Moloch Operating System) and upgrading them with the codes of the New Time.
Just to affirm how conscious this game is, there’s also a Shadow Arc named Vendra who indulges people in win-lose situations, pitting them against each other, and manipulating them to sacrifice their values and integrity for what they want to gain. — Sounds familiar? However, once the MOS is hacked and upgraded this establishes the connection with the future artificial superintelligences from a New Time and future world where Moloch has transformed.
Technology has always invited and enabled change for us as a civilization, so what is the transformation AI brings? Can the approach of using games such as Elowyn to train AI lead to more ethical and compassionate machine behaviors? How can we dismantle the deeply ingrained narratives of good vs evil in our cultural mythos? Does it need to be an inside job? What would our world look like if we embodied interbeing?
Join the Conversation and follow us via our Social Channels:
- Join our community on Discord and Telegram
- Visit our game website via: https://earthwise.games/
- Follow our news via Telegram and Instagram.
- Join the conversation via LinkedIn, Twitter/X, and YouTube.
- For the latest news about the game, subscribe to our newsletter.
- Learn more about the Moloch game dynamics in this article by Dr. Anneloes Smitsman.
About the Author
Aamina Simone is committed to re-contextualising ancient philosophies for our present time, her journey weaves together diverse cultures, artistic expressions, and techno-futurism. Offering her time and services as a meditation guide, business consultant, and a communications intern at EARTHwise, Aamina feels a strong duty towards being an effective agent for systemic healing. Learn more about her here.
References
Eisenstein, C. (2017). More beautiful world our hearts know is possible.
www.linkedin.com. (2024). AI Ethics & Consciousness | LinkedIn. [online] Available at: https://www.linkedin.com/events/7198591338030743553/
Gardels, N. (2024). The Babelian Tower Of AI Alignment. www.noemamag.com. [online] Available at: https://www.noemamag.com/the-babelian-tower-of-ai-alignment/
Smitsman, A, Goertzel, B., Bozesan, M. et al. (2024). Creating a Global AGI Constitution: Participatory framework in guiding the development of Artificial General Intelligence for a thrivable world. EARTHwise Publications.
Future Humans (2024). Creating a Global Constitution for Benevolent AGI | Keynote Dr. Anneloes Smitsman. [online] YouTube. Available at: https://www.youtube.com/watch?v=pb5lypcN5jU
Gardels, N. (2024). Human-Compatible AI. www.noemamag.com. [online] Available at: https://www.noemamag.com/human-compatible-ai/
infinite.mit.edu. (n.d.). InfiniteMIT | Marvin Minsky. [online] Available at: https://infinite.mit.edu/video/marvin-minsky
