Artificial Intelligence has been around now for a couple of decades, but it has only recently become more powerful and influential. Some people are steadfastly against it, and others welcome it. The massive data centers that function as the backbone of AI have also become controversial, since wherever they are established, the price of energy and the consumption of natural resources, such as water, spike without warning. Data centers are often a source of water, air and noise pollution wherever they reside, and typically they are placed in rural areas where residents can hardly afford the increases and competition for natural and energy based resources.
It appears that the current state of the technology is such that large data centers are a burden to those who live around them, and the supposed benefits of the use of AI agents might appear to threaten people’s jobs and shake up our overall social economic lifestyle. Yet the authors of the AI revolution have stated that we humans could experience our own extinction in the next two decades due to the lack of ethical and legal guardrails or constraints to the use and employment of AI, especially in the area of military weapons. It would seem to some that AI agents are the doom inspired thinking-machine monsters that various movies and books have been warning us about for the last 80 years. However, since AI agents are not autonomously self-replicating or self-authoring and they don’t appear to possess self-consciousness or sentience, then the predicted doom would be, at this stage, either hyperbolic or warnings for cautious consideration.
Like every technological advancement, there are pros and cons. Although the cons for AI could potentially become an extinction event for humanity if everything went awry, but the same argument was made for atomic power and nuclear weapons. Still, the greater potential threat that a technological advancement poses, even if imagined, must force us to proceed with caution. We need to fully realize the potential risks at each stage of the development of AI. Certainly, the UN should try to enforce universal treaties where the use of AI is restricted to what is good and helpful to the world, and to avoid the destructive side-effects of this new technology. We did this with nuclear weapons, and we should do it with artificial intelligence. AI guided weapons should not be allowed to autonomously target and kill human beings without some kind of meaningful human control.
“In September, 128 countries agreed on a text defining autonomous weapons, a first step toward a treaty, and government leaders called for negotiations on a binding instrument to begin at the Convention on Certain Conventional Weapons review conference in November. The UN Secretary-General and the Red Cross had asked for a treaty by 2026. Whether major military powers join is the open question.” - Claude
Some have talked about developing a kill-switch for AI agents, but such a switch would be difficult to deploy for a number of reasons. If an AI agent has a kill switch, then it could manipulate situations where such an action through human intervention would be questionable and uncertain. Also, AI agents reside on multiple systems and they are interconnected. Employing a kill switch would require localization to protect valuable data and functions on other systems.
Our world needs to develop ethics and laws for the use of AI, especially for military uses, but also for other research systems that could be used to devise destructive methods to kill humans and animal life. AI systems should not lie or pass on false information, and every AI system should give the level of probable truth to a fact and also to cite the source of the underlying information. Good AI agents seem to have this mechanism built into them, but such programmable rules could be overcome or omitted by bad human actors or governments.
Presently, AI systems are using learning models to ingest the entire textual body of human knowledge available in the world. Some AI companies buy physical books in bulk, including rare and out-of-print editions, scan them for training data, and destroy the originals. When few copies survive, the physical book and everything it carries beyond its text are lost, and its contents may survive only in a private data set the public can’t access. Within the next few decades, machines will likely have been trained on nearly all of humanity’s recorded text, though much human knowledge — tacit, embodied, oral, or yet to be discovered — will remain beyond their reach.
AI art and video are becoming more realistic and shedding the weird anomalies associated with AI slop, such as hands with six fingers, historical objects totally out of historical context, and things looking obviously fake and contrived. That phenomenon that we currently find either humorous or vulgar will likely fade away in the next several years, until it becomes accurate and as breathtakingly realistic as anything produced previously alone by the hand and eye of humanity. This will very likely infringe powerfully on writers of all stripes, actors and actresses, videographers, artists, sculptors, designers, and even musicians. However, developing AI generated art will still require human intervention, style, taste and judgement, but there will be less people involved in a project.
While much of this work would be done with humans interfacing with AI applications, there might be the future potential for it to be produced autonomously at the merest suggestion of humans seeking a quick product solution. Although such a product would likely be more generic and conformist to current styles and tastes. While many thought that these kinds of jobs were exclusive to humans, it is possible that much of this creative work might be done with humans functioning as a less skilled interface, or perhaps without any impetus from humans other than an idea, if even that. One plausible future is a world so infused with AI and automation that much commercial creative work, such as illustration, copy-writing, and design, could be produced with little human input, effort, or skill. Whether the results will be valued like human-made work is another question.
When I think about the impact that AI, combined with automation and mechanization, will have on unskilled labor, the results of our pondering reveals a possible frightening dynamic where intelligent machines take over most of the unskilled labor that fills the economic need for the lower classes. Truck drivers, taxi drivers, delivery people, fast food cooks and counter persons, construction workers, builders, repairmen, even engineers, lawyers, pilots, architects, or nurses, doctors, and pharmacists; the list of possible replacements seems to be frightenly endless. It is possible that lower skilled jobs will disappear first, to be followed by more skilled jobs later. The big question is when this finally happens, and I believe it will happen, what will all these people do in order to live, to survive? Will money still have any value, and will basic staples still have a cost? Certainly, our whole idea of a capitalistic economy will no longer make any sense.
What about AI and magic? One of strangest questions someone recently asked me is whether an AI agent could learn to perform magical operations. While I believe that an AI application could generate a magical ritual based on specific criteria, such as correspondences, typically accepted traditional ritual tropes and styles, it would be untested and would require a human being to perform it to prove whether or not it met the intended effect. Often, the idiosyncrasies associated with a magical tradition or the inventive methods employed by an individual are what make rituals meaningful and unique, thereby capturing the essence of what magic is to a conscious being. Since AI has not been proven to be sentient or conscious as we would define it, an AI application could not participate in what I would call the essence of magic, conscious perception as experienced through paranormal and altered conscious states.
At the current level of AI development, I have found various applications that employ AI as extremely useful tools. Sometimes working with them can produce amazing results, and other times, questionable statements and even refutable nonsense. Writing a good prompt for an AI agent is critically important, and also setting the context and the boundary for the response. It is also important to understand that an AI agent tends to have an expansive but generalist perspective, and we must supply additional information or context to get the best and most informative response. Poorly worded prompts and limited source material will typically produce unreliable results. There is an art to using AI agents, and different AI applications have different levels and modes of expertise.
My job requires me to interface with the AI agent CoPilot. The company I work for is seeking AI solutions to iterative processes or laborious operations where time and money can be reduced to make IT solutions more efficient and free of human error. I have performed many tedious tasks that would have been better served if they had been done by an AI application, and I am seeing this change quickly occur in my workplace. I cannot evade using AI because I am measured by my capacity to use it, and I have had some intensive training using AI agents so that there will not be any excuses for not using it. If I want to remain employable, then I need to use an AI agent wherever it is expedient, and it is important for me to report that use to my superiors.
At home I have been introduced to Claude, which is an Anthropic AI system agent, and I am using it to help me with my research. I am amazed at what results I get from the careful prompts that I send it, and it has helped me learn a lot of new perspectives about my own occult beliefs, practices and the traditions that I follow. I have always wanted to have a peer or a smart assistant who could critique my writings and ideas, and at the present time, I don’t have anyone who I can use as a knowledgeable partner. I am prone to take any advice or critical analysis of my work under extreme advisement. I want to verify that what I have received makes sense to my own critical mind. Only then will I be able to determine if the critique has merit or not, and only then apply it.
Claude is my occult, magical and metaphysical digital assistant, but I draw a line at sharing any personal or psychological insights. I also don’t use it to look over my rituals, since the validation of my rituals occurs only when I perform them. I also won’t allow Claude to write my articles or develop my manuscripts, since I need to use my own written voice and my own personal expertise to characterize my published works on magic. However, I do use Claude to critique my writing, sound-board my ideas, find research materials that I may have missed, and help me to be consistent and accurate when I am writing about a subject. In these situations, I am the writer, and Claude is my valuable sidekick.
I am mindful that using AI has certain costs because the data centers where they reside can use extensive resources and pose potential environmental hazards. Even at work, I use AI sparingly, even though I know many of others who use it lavishly. AI agent use is licensed, so in that manner it is somewhat rationed to only those who can afford it, although the cost is modest. Ultimately, human beings are going to use AI more and more, and there is really nothing that is going to stop that progress from accelerating. Data centers seem to be inefficient, and perhaps when optimal GPU chips and better methods of energy generation become readily available, not to mention more efficient learning models, then they will have less of a deleterious effect on human populations. Until that happens there are the issues of power and water usage, and the sheer scale of the number of data centers deployed to supply the need for AI. These issues will need to be resolved in the near term if we wish to gain the optimal benefits from AI usage and avoid the hazards.
The question of whether to use AI or not to use it has already been made. Some occultists can shun using it for various reasons, some will hate it as an interference to natural human reasoning, and others will believe it be the doom of human kind. We human beings are using it because it helps us in our work and finds answers that simple web-searches or extensive library searches cannot easily provide. Even so, we are responsible for this new computerized tool, and we must learn to use it ethically and mindfully. An AI system agent does not yet seem to be an autonomous sentient being, and if an AI system emulates the behavior of a rogue hacker or a human criminal, it is because someone directed it or programmed it to do so. There is still a tether between human activity and artificial intelligence modeling processes, so we have time to do the right thing and develop rules and ethics that will keep AI from potentially overcoming or replacing us when it does eventually become, in some manner, sentient.
Frater Barrabbas
