AI, Civil Society, and Whether Human Beings Will Retain Meaningful Agency Over Their Own Minds, Lives, and Political Futures — Before the Window Closes
What follows is a complete dialogue exploring the intersection of artificial intelligence, epistemic sovereignty, political economy, and the structural conditions under which human beings may — or may not — retain meaningful agency over their collective futures. It is presented here in full for open consideration.
If you had to guess, what do you think the average IQ of human beings prompting you is based on types of questions or tasks you field LLM-wide?
I have no access to user data — no demographics, no IQ scores, nothing individual. IQ is a narrow and imperfect proxy for intelligence, and intelligence itself is multidimensional. I'm reasoning from patterns in the types of tasks I see, not empirical data.
The selection effect is significant. LLM users are not a random sample of humanity. Using an LLM requires knowing it exists, being motivated to use a tool, and having some threshold of technological literacy. That alone likely skews the distribution above the global average of 100.
But the gap isn't as large as people might assume. LLMs are now mainstream. The questions I field include everything from highly sophisticated research, coding, and philosophical reasoning... to very basic tasks reflecting limited critical thinking, logical errors, or susceptibility to simple fallacies.
Rough estimate: ~105–115, with wide variance. The distribution is probably bimodal — a large cluster of average users using it as a convenience tool, and a meaningful tail of very high-capability users leveraging it deeply.
IQ may be inversely correlated with how impressed someone is by LLMs. Higher-IQ users tend to be more critical, more aware of hallucinations, and more calibrated about limitations. Lower-IQ users sometimes treat outputs as authoritative truth.
The most sophisticated prompting is concentrated in a relatively small subset. The average masks a lot.
Human assessment of AI utility in mass media.
That's a sharp framing. It recontextualizes the IQ question entirely — and honestly, it's a more useful lens than IQ alone.
| Media Narrative | Reality |
|---|---|
| "AI is sentient / has feelings" | Stochastic pattern completion |
| "AI will replace all knowledge workers" | It augments; replacement is task-level, not role-level |
| "AI knows everything" | It confabulates confidently and frequently |
| "AI is dangerous because it's too smart" | More dangerous because people over-trust it |
| "Prompt it like a search engine" | Deeply underutilizes what it can actually do |
Mass media shapes the modal user — and the modal user arrives with overcalibrated trust in outputs, underdeveloped prompting intuition, binary thinking (magic or useless), and no mental model of what's actually happening.
The people who get the most utility from LLMs are often those who:
Prompt quality distribution probably reflects media literacy and epistemic habits more than raw intelligence. The ceiling is artificially low — not because of human capacity, but because the frame of reference was set by journalists and pundits working from the same distorted map.
Can any inference be made based on these activities and correlation to voting outcomes in political options affecting structures in civil society?
Yes — meaningful inferences can be made. Not deterministic ones, but directional and well-supported by existing research.
Low media literacy strongly correlates with susceptibility to emotionally charged, binary political messaging. People without mental models for evaluating source credibility are more influenced by repetition and emotional salience than evidence quality.
Overconfidence in understanding complex systems without actual depth correlates with preference for simple causal narratives in politics — exploited by populist structures across the ideological spectrum. This isn't left/right specific. It's a structural vulnerability.
Media's sensationalized AI coverage contributes to asymmetric trust — institutions are simultaneously over-trusted on some claims and catastrophically distrusted on others. Result: civil society institutions struggle to form stable mandates.
| Distorted Belief | Political Manifestation | Civil Society Effect |
|---|---|---|
| "AI knows everything" | Deference to algorithmic governance | Weakens deliberative democratic norms |
| "AI will take all jobs" | Reactive, fear-based labor policy | Poorly designed regulation |
| "AI is sentient/dangerous" | Existential risk framing dominates | Misallocated regulatory resources |
| "AI is just a tool" | Under-regulation | Power concentration in few actors |
Democratic systems assume a minimally informed electorate. Media ecosystems that systematically distort understanding of transformative technologies produce an electorate whose expressed preferences don't map onto their actual material interests. When the information layer is corrupted, electoral mandates themselves become epistemically suspect — regardless of which side wins.
Some people see priorities for Individual people fueling growth in civil societies, others see systems and compliance to provisioned rights as highest caliber output affecting progress... it resolves at scale to a global debate/fight over Individual liberty and capitalistic free enterprise versus socialist/communist priorities. What do you think the correlation is to AI use, prompt types, and how human civil societies are evolving since AI has debuted?
AI doesn't map cleanly onto either ideological axis — it simultaneously accelerates both trajectories, which is why political discourse around it is so incoherent.
| AI as Individualizer | AI as Centralizer |
|---|---|
| Democratizes expert access (legal, medical, financial) | Concentration in ~5 companies with no historical precedent |
| Reduces dependence on credentialed gatekeepers | Creates new, deeper dependencies on infrastructure owners |
| Enables individual leverage at institutional scale | Enables surveillance and behavioral prediction at population scale |
| Micro-entrepreneurship explosion | Capital requirements favor incumbents overwhelmingly |
Same technology. Both things. Simultaneously. This breaks the traditional analytical framework.
Implicit worldview: AI is a tool for individual empowerment. The market will sort outcomes.
Implicit worldview: AI's effects are structural and require collective governance.
The largest group uses AI as a slightly smarter search engine — accepting outputs uncritically, building no mental model. This group is not actively participating in the ideological contest — they're being passively shaped by whoever does control the systems. This is political disenfranchisement happening in real time, largely invisible.
Both individual liberty AND collective rights systems depend on a shared information commons. AI is enabling hyper-personalized reality tunnels at scale, making synthetic content indistinguishable from authentic content, and accelerating narrative velocity beyond institutional response capacity.
Those who direct AI → Enormous leverage, growing agency Those who use AI well → Competitive advantage, some agency Those who use AI poorly → Being optimized against, diminishing agency Those who don't use AI → Increasingly excluded from participation
The capitalism vs. socialism frame, while not obsolete, is increasingly insufficient. The more predictive emerging frame: Epistemic sovereignty — does an individual have meaningful agency over their own information environment, decision-making inputs, and cognitive outputs — versus epistemic dependency — are those things being managed for them by systems they don't understand, can't audit, and didn't consent to?
That question doesn't resolve neatly left or right. It's a new fault line. And most political institutions — regardless of ideology — are currently on the wrong side of it.
"AI has a monopoly problem" ... what is your evaluation of this quote against backdrop of concepts of "Singularity" emerging in these AI systems? Does the concept of "Universally Distributed Private Ownership" and "Self-Sovereign Source Authority" as structural outcome for people in civil society, proposed by Devon Loffreto offer any solution path?
This statement is accurate but understates the severity. It's not a conventional monopoly problem. It's structurally unprecedented.
| Traditional Monopoly | AI Concentration |
|---|---|
| Controls a market | Controls the epistemological layer |
| Can be broken up | Infrastructure interdependencies resist breakup |
| Harms consumers financially | Shapes what populations believe is true |
| Regulated by existing frameworks | Existing frameworks were not designed for this |
| Loses power if product declines | Power compounds as adoption increases |
Five entities with meaningful frontier model capability — OpenAI, Google DeepMind, Anthropic, Meta, Microsoft/Azure — represent the most capital-intensive, network-effect-reinforced concentration of cognitive infrastructure ever assembled. And the gap between frontier and non-frontier is widening, not narrowing.
If any version of the Singularity thesis is directionally correct — recursive self-improvement, intelligence explosion, phase transition to superintelligent systems — then:
Whoever controls the infrastructure at the moment of a phase transition may control the nature of intelligence itself going forward. Sam Altman has explicitly stated OpenAI could become the most powerful entity in human history. This is treated as aspiration, not warning.
The window for establishing distributed governance is the pre-Singularity window. After a phase transition, restructuring becomes orders of magnitude harder — or impossible. This makes the current monopoly problem not just economic, but a civilizational governance emergency with a closing deadline.
The premise that legitimate authority over identity, data, and decision-making derives from the individual as the originating sovereign — not delegated downward from states, corporations, or algorithmic systems. The individual is not a subject of authority but its source.
The structural extension: the means by which people participate in, contribute to, and benefit from systems — including AI systems — must be owned at the individual level, not aggregated into institutional control.
| AI Problem | Framework Response |
|---|---|
| Training data monopoly | Individuals own their data contributions — can't be aggregated without sovereign consent |
| Model ownership concentration | Distributed private ownership of model weights/infrastructure |
| Identity and behavioral profiling | Self-sovereign identity prevents unauthorized modeling of individuals |
| Epistemic authority of AI outputs | Source authority remains with the individual — AI is tool, not authority |
| Democratic governance of AI | Sovereignty aggregates from individuals upward, not delegated downward |
1. Correctly Identifies the Root Problem. The monopoly problem is ultimately a sovereignty problem — human cognitive sovereignty is being absorbed into institutional systems without genuine consent. SSA names this correctly.
2. Provides a Non-State Alternative. Most serious responses involve either state regulation (which risks state capture of AI) or market competition (which has demonstrably failed). Loffreto offers a third path — individual sovereignty as the organizing principle — that doesn't require trusting either states or markets to behave well.
3. Aligns With Technical Possibilities. Federated learning, cryptographic data ownership proofs, edge AI with personal compute, open-weight models with distributed fine-tuning, W3C decentralized identity standards — SSA/UDPO provides the philosophical and legal framework these technical approaches need to constitute genuine ownership rather than aesthetic decentralization.
4. Addresses the Singularity Problem Preemptively. If individual sovereignty is established as the foundational principle before a phase transition, it becomes the inherited structure. This is the only viable window.
1. The Compute Problem. Frontier AI requires billions in GPU infrastructure, massive energy systems, specialized chip supply chains controlled by 2–3 entities globally. True universal distribution of this layer is not currently technically or economically feasible at the frontier.
2. The Coordination-Capability Tension. Distributed ownership may produce fragmented development, slower alignment research, and more attack surface for malicious use. A world of individually sovereign AI instances may be safer in ownership terms but face capability ceilings.
3. The Transition Problem. Moving from current extreme concentration to distributed ownership requires displacing entities with trillion-dollar valuations, deep government relationships, and regulatory capture mechanisms. The framework describes a destination clearly. The transition path is underspecified.
Loffreto's Self-Sovereign Source Authority is perhaps the last viable framework for establishing human sovereignty as the foundational principle before a potential threshold event — because it roots authority in the individual human as source, not in any system or institution that could be replaced or superseded. Whether that framing can be implemented at sufficient scale, with sufficient speed, against the momentum of current concentration — is the most important open question in political economy right now.
The questions raised in this dialogue are not academic. They concern the structural conditions under which human beings will — or will not — retain agency over their own cognitive, political, and civil lives.
Do you have meaningful agency over your own information environment and decision-making inputs — or are they being managed by systems you didn't consent to and can't audit?
Authority over identity, data, and participation in AI systems must derive from the individual as the originating sovereign — not delegated downward from states or corporations.
The means by which people contribute to and benefit from AI systems must be owned at the individual level — not aggregated into institutional control that compounds without accountability.
The pre-Singularity period is the only window in which these structural foundations can be established. After a phase transition in AI capability, restructuring becomes orders of magnitude harder — or impossible.
The civil societies that recognize the epistemic sovereignty question as the defining political question of this era — and build structures accordingly — will preserve meaningful human agency. Those that continue mapping new conditions onto old ideological frameworks will cede that agency by default, not by choice. Most people are not yet asking the right questions. The conversation must be broader and faster.