Seraphim (Bruce) Foltz: On the Promise and Perils of Artificial Intelligence
Seraphim (Bruce) Foltz is emeritus professor of philosophy at Eckerd College in St. Petersburg, Florida. Foltz draws on the history of philosophy (especially Ancient Greek, Byzantine, and Russian philosophy).
The technology of artificial Intelligence (AI) is exceptionally useful for innumerable tasks, not least of which is the rapid collection of information. It can serve as a ready almanac, a source of product reviews, a treasury of medical information, a vast compendium of encyclopedias of all kinds, and so on—but of course with the proviso that its answers be corroborated with more reliable (and less “robotic”) sources. And in combination with its many other capabilities—graphics, spreadsheets, and far more sophisticated applications—AI can actually be necessary if professional and business activities are to remain competitive. But it also carries with it a host of dangers, some of them extremely serious. This brief outline of several of these dangers will focus on AI chatbots (CBs), such as ChatGPT, Gemini and Grok that easily are freely accessible to the ordinary user.
1. Concealment of Bias: CBs typically present their results as authoritative and final, and this hides or disguises a range of biases, not only in the sources chosen but as programmed into the AI algorithms
2. Personal Impact: CBs can make us lazy and dependent on them, just as GPS navigation can cause our abilities for geographical orientation to atrophy. We forget how to exercise our own capacity for recollection, as well our ability to reason for ourselves.
3.Fragmentation: Gathered information is presented to the user outside of any context, and this tends to fragment knowledge itself as well as the worldview of the user.
4. Master-Slave Dialectic: First articulated by Hegel, and appropriated by Marx, the interaction of master and slave tends toward a reversal, in which the master (increasingly dependent on the slave) gradually becomes the slave of the formerly enslaved, who does the actual work. AI can make us increasingly dependent and indeed slavish.
5. Distraction: The pleasurable satisfaction of one curiosity after another can intensify the distraction already at work in social media and mass media, causing us to lose any sense of the gravity of our life on earth, our one opportunity to prepare for eternity. Indeed, St Aimilianos called “distraction” the greatest danger posed by modern technology.
6. Demonic Energies: We typically pose questions to the CBs, and any posing of questions opens up the space of a personal relation, even if it is asking a passerby for directions. This personal space requires varying degrees of faith in the one questioned, while interpretation of the answers requires reflection of the questioner upon himself. The one questioned may be an automobile mechanics, a marriage counselor, a confessor or spiritual father, or God Himself, with ascending degrees of both personal involvement and of spiritual danger. But unlike these examples, the CB that is personally interrogated is not a face or countenance turned toward us, but a metaphysical “black hole” into which we will invariably project our own imaginings. This makes this still-personal relation more like the kinds of divination forbidden by the Orthodox Church, precisely because this space is so rich with opportunities for the intrusion of demonic energies. And as the questions asked of the CB ascend from asking the date for the Fall of Constantinople to asking for the best place to vacation, from asking for marital advice to asking about the meaning of life, the “space” for demonic intrusion opens ever more dramatically. The subtle dangers of prelest or spiritual delusion, of dark and paranoid worldviews inviting preemptive force or violence, as well as various modes of demonic possession, cannot be underestimated.
7. Averting These Dangers. Most obviously, a robust and observant Orthodox life, with constant nepsis or mindfulness is the best safeguard against all these perils. But a more philosophical approach can also be helpful, such as Heidegger’s prescription of Gelassenheit or “releasement” toward technology, which he appropriated from the spiritual counsels of Meister Eckhart. For after all, AI is nothing more than the operation of (very sophisticated) machines, and only if we believe instead that AI represents some kind of higher power, and only if we invest too much of ourselves in it, are these dangers enabled. And this same caution would apply to the spiritual hyperventilation underlying the phobic warnings of AI “taking over” the world. Rather, a “calm” or “released” approach to technology, to AI, and to our overall worldview, is to simply let this technology (and all others) be what they are and nothing more, i.e. useful tools ultimately no different from a hammer or a rake, and thereby releasing them from our own fantastic expectations—while promoting a releasement of ourselves from dependence on them and from slavery to powers far darker than AI itself.
Katina Michael, M.G. Michael: AI, Machine Learning, LLMs: Benefits and Risks
Katina Michael is a professor and visiting research scientist at Arizona State University in the School for the Future of Innovation in Society and adjunct in the School of Computing and Augmented Intelligence. She has completed a Doctor of Philosophy, Master of Transnational Crime Prevention, and a Bachelor of Information Technology.
M.G. Michael is an Eastern Orthodox theologian. He was formerly an honorary Associate Professor at the University of Wollongong Australia in the School of Computing and Information Technology. He has completed a Doctor of Philosophy, Master of Arts (Hons), Master of Theology, Bachelor of Theology, and Bachelor of Arts.
Artificial intelligence (AI) was originally conceived in the 1950s. In its original conception AI was about mimicking human reasoning through logic and formal rules (e.g., using logic-based reasoning and rule-based expert systems). The AI systems before 2010 were largely rooted in formal logic, rational decision-making, and algorithmic derivation, before data-driven methods emerged, such as machine-learning (e.g., neural networks and deep learning) that are based on statistical approaches.
There are many benefits to the adoption of traditional AI. Algorithms can be used to follow formal rules and automate a great deal of decision-making by humans. This creates transparency based on logic that can be audited if required at a later date, holding organizations accountable for errors based on poorly constructed algorithms or wrong data inputs. Beyond increased automation, commentators point to efficiency gains, improved predictive analysis, science and technological discovery, better customer experience due to 24x7 availability, economic growth and innovation, and societal and environmental benefits, though the latter are debatable. Given modern AI systems use approaches based on statistics with large numbers of parameters, we really cannot guarantee a process of formal verification. Responsible AI is an emergent area of research that describes the development and deployment of AI systems in a way that is ethical, transparent, accountable, and aligned with human values. It ensures that AI technologies benefit individuals and society while minimizing harm. Some of the negatives of AI include: job displacement, bias and discrimination, loss of privacy and transparency, security issues, and ethical and social risks, the concentration of power and existential and long-term risks that are for the greater part unknown.
Enter large language models (LLMs). Rather than being explicitly programmed, LLMs are trained on large data sets and look for patterns in data (e.g., text, audio, images). It is important to note that LLMs are predictive, they are not meant to verify; they are more concerned with determining something is plausible and not necessarily concerned with accuracy. LLMs also do not have an in-built mechanism to distinguish between truth and lies. OpenAI’s ChatGPT-3 in 2020, for example, was built on 175 billion parameters. That number is estimated to be much higher today with GPT-4o in 2024. While with each version of GPT there are presumed advancements, greater numbers of parameters do not equate to quality, or even performance. While many are citing productivity increases as a result of the adoption of generative AI, recent studies point to a generation who will grow up without important skills such as critical thinking, basic language and arithmetic skills, the ability to discern a path forward when several options exist. Some are even citing the potential for loneliness and brain atrophy if one over-relies on GPTs, as opposed to seeking friendship from real humans, or trying to problem-solve.
If we teach a generation of young people that they should run to a machine for comfort, they will discover there the absence of actual warmth and care, even if the inanimate object looks and feels human. Again, ChatGPT noted of itself: “As an AI language model, I do not have feelings or emotions…” As Metropolitan Kallistos Ware emphasizes, “You may love your computer but your computer does not love you”. Robotics have so advanced to seem almost human, but we are and will be absolutely fooling the senses and the mind to believe that what stands before us is a truly sentient and feeling life form. This deception of ontological proportions, dealing directly with the nature of being, can only continue for so long before there is the inevitable breakdown in communication: “I’m sorry Dave, I’m afraid I can’t do that” and “Dave, this conversation can serve no purpose anymore. Goodbye”.
Ultimately, we maintain: God is not an AGI, if only by virtue of the Supreme Being’s immutability alone. The Supreme Being cannot be contained in any deep learning model nor be recreated in our own image and likeness, and surely not as a machine. People need human succor, and a compassionate embrace, they need real love, and not simulated. They need to hear counsel from a loving heart, which together with all else, understands doubt and is vulnerable in and of itself. The machine, no matter how sophisticated the algorithms, does not love, because it cannot love the way we alone can, it does not possess the same inspired spirit as their builder. And computers, regardless of how powerful they are, they are limited and not limitless. God is without beginning or end, computers are made up of 1s and 0s.