Artificial Intelligence is arguably the most important, disruptive technological invention in human history. At the same time, AI acts as the perfect narrative cover for the largest power grab ever attempted by humans. These truths are not at odds with one another, and being able to intellectually navigate both simultaneously and successfully requires a nuanced framework that has, in my opinion, failed to materialize thus far in the public conversation.
The very idea of making an Artificial Intelligence lies within the combination of the intangible with the tangible –– the conscious with the unconscious, the metaphysical with the physical, the spiritual with the worldly, the psychological with the physiological, etc. Today’s AI programmers and chip manufacturers are yesterday’s alchemists –– attempting to turn sand into thought in lieu of lead into gold.
An issue I find in discussing AI is that the discourse so quickly sours to straight up propaganda from its creators or –– perhaps more insidiously –– surface-level, esoteric “mumbo jumbo” (see above) that muddies the ability to create informed consent for the masses. I have yet to resonate fully with any would-be sides of any would-be debate on the matter and remain quite skeptical of the public’s perception on its implications today, not to mention tomorrow. In my opinion, no champion of the technology has properly addressed the immediate and long-term effects from this distributed state-change of generative and analytical compute.
These two axioms –– tangibility and time –– I intend to use as scaffolding to house my thoughts, which admittedly contain much conflict, confusion, and unknown ignorance. Before diving in, I’d like to preface this purposefully “negative” opinion piece about the downsides with a brief caveat that I do believe there are a handful of use cases for this technology that are quite interesting as it relates to the supposed goal of “empowering the individual.” That said, I think the pros are far outweighed by the cons, and thus –– in addition to my belief that there are far more pro-AI talking points circulating than negative –– I will spare you from such and focus on the less discussed downsides of this disruptive technology.
The Tangible Today: Big Data and Surveillance
The most tangible threat vectors brought about by today’s proliferation of LLMs relate to data –– both the personal data willingly given by its users, and the immense amount of data already amassed. Both of these venues for surveillance, not to mention algorithmic-driven predictive policing, are enabled entirely by the incredible ability of AI to comb through ungodly amounts of data with ease, and to efficiently heuristically analyze people and things into a network from an otherwise insurmountable cache of data. While many proponents of the technology counter this notion with the suggestion of using locally-run, open source models to protect your data, this ultimately does little to truly minimize the effects of a terminally-shrinking anonymity set for internet users.
Unfortunately, simply opting out of using closed source AI does not negate the state change to world surveillance brought about by the technology. Choosing not to put a Ring camera on your front door matters little if your neighbor across the street has one facing your house. Furthermore, even if you convince your neighbor to take his camera down, and convince the greater neighborhood to remove their Flock-style cameras, AI-enabled surveillance satellites passing overhead and driver-less electric vehicles full of sensors passing by shrink your “meat-space” anonymity sets significantly.
The internet is an open network that communicates between nodes by breaking down larger data sets into tiny, easily transmissible packets, which are reconstructed into said larger data sets once they arrive at their final destination. For example, a video call you make on your cellphone is beamed to your conversational counterpart in compartmentalized bits of data via hardware infrastructure owned and operated by internet service providers. These packets are often broadcast without any sort of encryption, and said ISPs or software providers such as Google or Facebook, can easily amass and store immense data tranches of user activity, which up until now was a relatively useless venture due to the sheer volume of internet data volume. These same tools and data sets are now available to domestic and foreign intelligence, to domestic and foreign AI companies, and to domestic and foreign military powers, law enforcement, and court systems. Unfortunately, no longer can our activities online find safety within numbers.
As more and more humans become reliant on LLMs for task managing, communications, e-commerce, content creation, and even investment management, the amount of personal data that becomes beholden to these often privately-owned companies and their user agreements grows in magnitude –– often just one click away from being entered in the State’s data coffers. As more user data is accumulated in various silos, the want to forcibly take said data increases, and with AI-enabled exploits and advanced phishing techniques (see Revolut’s latest whoopsie) the ability to keep it safe is rapidly decreasing. AI necessitates more data to train on while also encouraging more user data be provided to said data silos.
This issue is further exacerbated due to a self-enforcing loop of human reliance on these very systems –– more usage today means more prompts and user data sent to the training sets, which in turn are used to strengthen and maximize the “stickiness” of users. Unfortunately, this reliance loop of data provided and data stored has likely reached some semblance of escape velocity, and trying to shove this system back in the proverbial Pandora’s box will fail on any level other than the extremely local level –– e.g. an individual opting out within their own home. Ultimately, this does nothing to address the very real state-change of surveillance enabled by AI in the real world.
As it stands today, the world at large has seemingly chosen to keep feeding this metaphorical runaway train with data centers, kilowatts, and dollars, blurring the lines between the public-private partnerships upholding this new industry. Many of the fears of a State-controlled, Big Brother system are only feasible with an enforceable, bureaucracy-scaling AI –– whether the system is ultimately co-opted by the State or run by a small, private group of technologists from Northern California is irrelevant.
All that matters is that it exists, and that it can be utilized by the humans hiding behind its linear algebraic curtains.
The Tangible Tomorrow: Empowering The Breakaway
One of the greatest tricks the AI industry deploys at large is focusing the discourse entirely on the technological level, and ignoring the human element. But on occasion, the idea of human empowerment from distributed intelligence is discussed on the individual level. For example, an entrepreneur’s small business is able to scale far more quickly due to deflationary efficiency gains from cutting costs on expenses surrounding human labor and time. This is generally viewed in a positive light, and there are certainly merits to it on shorter time frames. Unfortunately, tomorrow’s end game to this deflationary axiom is quite centralizing, with efficiency gains for them, and unemployment lines for you. The exponentiality of this very real mechanism is knowingly distorted by those most poised to benefit from it –– the stewards, creators, investors, and legislators –– while the entirety of the focus stays on the short-term empowerment of the working class.
Very real and very tangible nexuses of power, whether at the State level or within the private sector, already exist. AI-enabled automation and analysis allows these very parties an opportunity to seize the means of production and centralize the profits and efficiency gains downstream of this deflationary technology. While most rhetoric involving AI deceleration is framed by AI accelerators as CCP propaganda or communist gobbledygook, how exactly is the working class empowered by a small minority dominating the fields of AI-enabled, automated manufacturing of AI-enabled drones and droids? The fields of warfare, healthcare, education, construction, finance, agriculture and even hospitality and the arts, are up for serious and permanent revolution moving forward from these advancements. What appears as disruption and decentralization today can quickly turn to destruction and consolidation tomorrow.
For example, AI’s so-called “distributed efficiency gains” are capped by the masses’ ability to generate compute, which in turn requires immense amounts of energy and chip fabrication capabilities. Unfortunately, energy capacity and production remains one of the most State-controlled industries on Earth, despite recent gains in solar and battery technology. In addition, chip fabrication might actually be the most centralized industry on Earth. As distributed compute reduces the choke points of control to power and fabrication, the plausibility of these industries somehow migrating out from under the thumb of its controllers to the benefit of AI users diminishes the further into tomorrow we go.
The very AI-driven empowerment that is touted by its stewards also applies equally to the already powerful –– be it the literal State, the private companies and investors, or the fraternities within the breakaway merger of public and private interests which has dominated our lives both politically and economically for many decades. Many of my personally held fears regarding the State’s empowerment from tomorrow’s digitized economy –– in spite of today’s personal gains from digital asset appreciation –– are similarly ignored by AI sycophants who refuse to acknowledge the parallels of centralized pockets of control being empowered by AI –– open source or otherwise. There is a further parallel between the two models, given that the warm fuzzy thoughts reserved for open source AI are similar to those showered upon privately-issued dollar stablecoins, while closed source AI plays the role of CBDCs with respect to the surveillable and programmable money debate. (Even their marketing lingo has shocking similarities –– “braining the unbrained” is the new “banking the unbanked.”)
Compounding on this unfortunate reality is the idea that as the means of production –– be it for power capacity, chip fabrication, or compute –– are gobbled up by this breakaway power’s use and control of AI-enabled manufacturing, our would-be dissenters are stuck in group thought, which itself is distributed and homogenized by continued LLM use.
The Intangible Today: Viral Replication of Homogenized Thought
While we have discussed AI’s ability to analyze larges swathes of data, little time thus far has been given to its ability to generate content. This is likely due to the fact that today’s effects from AI’s generative capabilities are relatively intangible upon first glance. For starters, thought generation and content distribution, as it stands today, have centralized discourse to be native to social media platforms and trained on shockingly similar datasets –– both of which are determined by a small set of Silicon Valley stalwarts. This means that what you believe to be an efficiency gain as it relates to content creation is in fact an homogenization of thought, distributed on rails patrolled by curative bots and algorithms. This combination and context exacerbates a viral-esque replication of human homogeneity, with numerous and intangible consequences for today.
For example, this coagulation of content has huge effects on the ability to organize and properly discuss whether humans should actually be accelerating this at the scale that is currently being pushed today. Quite frankly, there is a case to be made that the entire world economy has been tethered to this idea that AI will “work” and that any debt taken on today will be offset with said efficiency gains and deflationary effects. Unfortunately, the very nature of AI’s generative and distributive nature, paired with a suite of malleable Silicon Valley-created content algorithms –– not to mention the monetary incentives at the nation and influencer-level –– has created the worst possible environment for appropriate discourse on the technology itself. Thus, it is my suggestion that you do your best to ignore the constructed and simulated mass fear of being left behind that plagues the purchased propaganda permeating today. It is okay to question the narratives of the Techno-Optimists, many of whom have merged their private tech firms with the National Security State. It is more than reasonable to question the motives of the AI industry. The accelerationists claim they are pushing the frontier of human thought with their research, but then shame those thinking outside the regulated lanes –– for thinking cautiously.
When we look back, I fear we will realize that many of the compounding mistakes made due to ignorance of the long-term effects of AI use could have been avoided with proper care regarding the unknowns of today.
The Intangible Tomorrow: The Suffocated Spirit
The last quadrant, the intangible tomorrow, unfortunately necessitates some discussion of the aforementioned esoteric “mumbo jumbo.” The generative, creative force of nature –– whether internally in the mind of an individual human or in a larger, physical ecosystem –– is in serious danger. Both individuals and collective humanity are at risk from the onslaught of slop corrupting humanity’s own intangible “training sets” as it dominates our most prominent cultural battlegrounds. Much like how an AI’s training set can become corrupted by training on its own generative content, so, too, can the prompter’s knowledge base become tainted from their incessant use of generative tools in lieu of their own intellect and creativity. Furthermore, even those that choose not to prompt a machine when, for example, writing an op-ed, their constant exposure to slop –– be it articles, podcast scripts, memes, music, or less obvious slop-portunities such as take-out menus, logos or websites –– will still influence their inspiria.
Much like how moving to a new country that speaks a new language can change your unconscious mind and affect your dreams, humanity moving into a new, AI-driven world with a new generative language for rhetoric and design will certainly have an intangible effect on our unconscious minds, be it collectively or individually. Not to vouch for the rest of their philosophy, which I certainly do not, but one could borrow the Theosophic concept of the “akashic records” as a parallel to the training sets that LLMs parse when they generate content. As more and more of the recent end of the “akashic record” –– the collective experience of humanity –– is dominated by experiences that are influenced and generated by artificial intelligence, concurrently less and less of the overall human experience remains human-derived. The long-term effects from this are simply unknown.
The off-shoring of one’s cognitive output to a LLM is one thing, but the communal consequences of society and culture doing so could have immense implications for the intangible –– for creativity, for the soul, for the generative nature of life itself. To the latter point, the very start of existence itself –– something from nothing –– has led to multitudes of spiritual, philosophical and religious reasoning throughout human history. The only certainty of such an event is that no one was there to observe, and while I do not fully understand my own reasoning, I fear the something-from-nothing nature of life itself could be buried by the amassing of surveillance data and persistent context enabled by AI.
Life and human creativity becomes buried in slop, while culture and society are suffocated by persistent surveillance –– regardless of AGI or ASI being achieved.
Yesterday and The AGI Paradox
I think there is a not-so great reason why the discourse up until now has focused so much on whether or not AGI is achieved, or if that is even possible –– it becomes a semantics issue, and it ultimately does not matter. The great fear around AI today, p(Doom) or the existential threat from a rogue artificial generative intelligence (AGI), fails to contain the reality that occurs regardless of a Skynet-esque act of war from our creation.
In my opinion, AI doom and AI abundance are yet another false binary, and that humanity will forever change with the onset and proliferation of modern AI algorithms, regardless if it meets the requirements for achieving AGI. To put it «poetically,” an algorithm need not contemplate existence in order to snuff it out. Hyper focus on the very unlikely extremes –– p(Doom) or AGI-enabled utopia –– clouds today’s discourse from discussing not only what has already happened due to AI, but what will likely happen.
I am less afraid for humanity’s doomsday caused by AI than I am of humanity doom-scrolling itself into irrelevance and spirit death caused by AI. I certainly do not think a top-down ban on the technology will do any good but further centralize control on an already centralizing technology, but I am more hopeful that a bottom-up resistance to AI could help us preserve humanity’s spirit –– warts and all.
After sharing an early version of this piece with a few trusted friends with very pro-AI stances, an opinion I kept hearing was that if the surveillance state was already so entrenched alongside the negatives outline above, why not just use the technology today? Unfortunately, I am unable to come to terms with a few elements of such a line of thinking.
For starters, I believe strongly that not every axiom of privacy is a dead end, and I will not gate-keep nor “doompill” about the very real efforts being put into carving out privacy-enabled enclaves, digital or otherwise, for humans to utilize. Secondly, I also believe strongly that the world –– or rather, the elite’s perception of the world –– relies on a karmic law of some innately unknowable sort. Having spent a handful of years now immersed in the dissection of the-powers-that-be’s plans, it seems apparent to me that revelation of the method is of utmost importance to them, and there are numerous iterations of hidden symbolism and “quiet part out loud” quotes from said leaders and their henchmen. For these reasons –– as ridiculous as they may seem to some –– I will not consent to giving up the privacy front in order to slightly streamline some processes in my day-to-day with AI.
Instead, I urge those that wish to normalize discussing an aversion to AI-generated content. Better yet, go out of your way to support AI-free content! Carefully curate your and your families’ intake of the aforementioned slop, and remain a steward of all that was achieved by humans in the days before silicon could think for us. Humanity is certainly barreling towards a new frontier –– one with many, many stated promises, and a few less-spoken fears. AI will certainly bring about a new world, just maybe not the one we were sold.
In order to ensure our tomorrow, all we must do today is remember yesterday.
