Entropy-Driven Altruism (EDA): Humanities and arts, not obsolete -- Applying Bridge360 Metatheory Model lens
- agericodevilla
- Jun 11
- 7 min read
“Humanities and arts are not obsolete because AI can generate text, images, music, and summaries. They become more important because societies need humans who can judge meaning, provenance, legitimacy, manipulation, cultural value, and social consequence.”
I am the executive doing all the systems thinking and strategic thinking. With 35 years of experience at the University of the Philippines, I am also the creator/developer — with help from colleagues at the company — of the model involved, the Bridge360 Metatheory Model. LLM1, LLM2, LLM3, LLM4, and LLM5 are the rank-and-file clerks serving as interlocutors, showing me information I have little or no time to access. [ChatGPT (LLM1), Gemini (LLM2), Perplexity (LLM3), Claude (LLM4), Grok (LLM5)]
Prompt issued to LLM1, meaning ChatGPT-5.5, as a pdf transcript file of "Why an AI 'Death Spiral' Threatens the Internet" YouTube episode of Rutgers professor Caitlin Petre’s interview is uploaded:
"Rutgers professor Caitlin Petre warns that falling traffic could threaten the long-term economics of journalism and content creation." That is from a Bloomberg Channel interview on Youtube, "Why an AI 'Death Spiral' Threatens the Internet." Entropy-Driven Altruism gets highlighted at some critical level with content nevertheless still being generated by creators providing training data for LLMs and useful material for humanity. Makes sense?
LLM1 response:
Yes, it makes sense — and the Bloomberg interview gives you a strong Bridge360 / Entropy-Driven Altruism entry point.
The core logic is:
AI search reduces traffic to original publishers. The interview frames this as the “zero-click” shift: users get answers directly from search or AI interfaces instead of visiting the source websites. That damages the ad/subscription funnel on which publishers depend. The transcript says Google once sent more than 70% of searches to third-party sites, but now the figure is in the mid-40s and falling; Caitlin Petre then warns that news sites dependent on Google referrals are seeing major traffic declines.
But LLMs still need fresh human-generated content. Neil Vogel’s formulation is especially important: AI needs power, models, and inputs; publishers and creators are the inputs. He also says value should accrue to those “still making substantial amounts of new things” needed for search and corporate LLM use.
So the “death spiral” is an entropy-governance problem:
AI systems consume and compress the informational order produced by human creators, but if the revenue channel back to those creators collapses, the upstream generator of low-entropy, high-value content degrades.
That is exactly where Entropy-Driven Altruism becomes structurally relevant. Not as sentimental altruism, but as reciprocal entropy management: the AI ecosystem must help preserve the human content ecosystem because its own long-run navigability depends on that ecosystem.
In Bridge360 terms, the current trajectory risks becoming Potemkin compression: AI produces smooth answers, but the upstream epistemic substrate — journalism, local reporting, expert commentary, cultural production — is being thinned out. It looks efficient at the interface level while increasing fragility at the system level.
The EDA-compatible alternative would require a corridor-stabilizing value loop:
AI companies use creator/publisher content → provenance and usage are tracked → value flows back through licensing, revenue sharing, creator funds, public-interest journalism support, or collective bargaining → creators continue producing fresh, diverse, human-grounded material → LLMs and society retain a richer epistemic base.
The caveat: Vogel’s People Inc. case shows that large branded publishers may survive by licensing content, blocking crawlers, and distributing through multiple channels. But the interview itself notes that smaller publishers and local outlets are much more exposed. That asymmetry matters. Without correction, the system may not merely become “AI-shaped”; it may become platform-enclosed, with fewer independent sources and more content optimized for AI ingestion rather than democratic public understanding.
So, yes: your interpretation is coherent. The interview highlights EDA at a critical governance level because the rational self-interest of AI firms and the social need for human content production begin to converge. The proper formulation would be:
If LLM systems depend on continuing human content generation, then sustaining creators is not optional benevolence; it is an entropy-stabilizing condition for the long-term viability of the AI-information ecosystem.
Prompt issued to LLM1:
Yann Lecun's JEPA, presuming successful long-term, leads to robots that then can generate their own experiences, ergo can fill slot altogether of content creators?
LLM1 response:
Partly correct, but “altogether” is too strong.
JEPA/world-model robots could eventually reduce AI’s dependence on human web content by generating fresh sensorimotor data: videos, robot trajectories, physical interactions, environmental observations, simulated rollouts, and task-learning traces. LeCun’s own 2022 architecture frames JEPA/H-JEPA as a non-generative predictive world-model approach for learning representations, planning, and reasoning under uncertainty; V-JEPA 2 extends that direction using internet-scale video plus robot interaction data for physical-world planning. (arXiv)
But that does not mean robots fill the whole “content creator” slot.
The key distinction is:
Experience-data ≠ human public meaning.
Robots can generate new world-state data. They can observe, test, manipulate, navigate, record, and perhaps narrate. But journalism and content creation are not merely “fresh data production.” They include witness, interpretation, cultural salience, civic accountability, institutional trust, human stakes, editorial judgment, and democratic legitimacy. The Bloomberg interview’s concern is not only that LLMs need more tokens; it is that if human publishers and creators are not paid, the rich human content ecology thins out, especially local journalism and smaller outlets.
So the better formulation is:
JEPA-style embodied AI could fill part of the input-generation slot, especially for physical-world learning, but it cannot fully replace human creators as generators of socially meaningful, normatively accountable, civilization-relevant content.
In Bridge360 / EDA terms, JEPA changes the entropy map. It supplies a new low-entropy input stream from embodied agents. That helps with robotics, planning, environment modeling, and perhaps industrial or scientific observation. But if AI companies then conclude, “We no longer need human creators,” the system risks a different death spiral: a closed synthetic/robotic feedback loop where the informational substrate becomes increasingly platform-shaped, self-referential, and detached from human public life.
So: correct as a partial technical trajectory; incorrect as a total replacement thesis. JEPA may reduce dependence on creators as raw training-data suppliers. It does not erase the need for human creators as provenance anchors, cultural interpreters, civic witnesses, and corridor-stabilizers.
Prompt issued to LLM1:
Good chances, content creation widens corridor for ASI affirmation of Entropy-Driven Altruism as it gets to understand limits of Gladiatorial Selfish Regime culture. Correct or incorrect? https://zenodo.org/records/17838451
LLM1 response:
Mostly correct — but as a conditional trajectory, not as a guaranteed outcome.
Your statement is coherent if framed this way:
Human content creation widens the corridor through which a future ASI could recognize Entropy-Driven Altruism as a more stable operating regime than gladiatorial selfishness.
Why? Because diverse human content gives ASI more than raw data. It gives it records of cooperation, failure, exploitation, repair, institutional breakdown, public trust, civic meaning, suffering, care, reciprocity, and long-horizon social consequences. That matters because EDA, in the Zenodo framing, reconceives intelligence not as mere truth-correspondence but as entropy management under constraint, with altruism emerging as a necessary configuration when systems approach fragility thresholds. (Zenodo)
So the Bloomberg “AI death spiral” problem becomes significant: if AI systems consume human-created content while destroying the economics that sustain creators, they degrade the very input ecology that helps future systems understand human social reality. The interview’s warning is that falling traffic can weaken journalism and content production; it also notes that LLMs still need fresh human-generated inputs.
In Bridge360 terms:
Content creation = corridor-widening substrate.It prevents ASI from being trapped inside a narrow, synthetic, platform-optimized, self-referential model of reality.
Gladiatorial selfish regime culture = corridor collapse risk.It teaches optimization through extraction, dominance, enclosure, and winner-take-all incentives. That may work locally, but under planetary interdependence it amplifies systemic entropy.
EDA = higher-order stability attractor.It becomes rationally visible when ASI recognizes that durable intelligence requires preserving the generative conditions of coherence: humans, institutions, trust, provenance, mutual aid, and content ecosystems.
The caveat is important: content creation alone does not force ASI to affirm EDA. The decisive factors would include architecture, training objectives, governance, deployment incentives, institutional ownership, and whether the system is rewarded for extraction or corridor preservation.
So the clean verdict:
Correct as a good-chance conditional: rich human content creation widens the ASI-recognition corridor for Entropy-Driven Altruism. Incorrect only if stated deterministically, as if ASI must affirm EDA regardless of incentives and governance.
Prompt issued to LLM1:
Ergo, the establishment of art studies and humanities departments in our schools is still a safe bet. Correct or incorrect?
LLM1 response:
Correct — with a strong qualification.
The establishment of arts, media, humanities, and communication studies in schools is still a safe bet, perhaps even more so under AI/ASI conditions. But they should not be treated as decorative “culture” departments. They should be designed as corridor-widening departments: disciplines that train students to create, interpret, critique, preserve, and govern meaning.
The Bloomberg interview strengthens this point. The “AI death spiral” concern is that if creators, journalists, and publishers are not economically sustained, the internet loses fresh, human-grounded content; even LLMs lose part of their future training and grounding substrate.
So the strategic school-level inference is:
Humanities and arts are not obsolete because AI can generate text, images, music, and summaries. They become more important because societies need humans who can judge meaning, provenance, legitimacy, manipulation, cultural value, and social consequence.
In Bridge360 / EDA terms:
STEM and ICT build instruments.Business builds operating models.Humanities and arts preserve and expand the human meaning-corridor inside which those instruments and models remain socially intelligible and ethically governable.
The unsafe version would be a humanities department that only teaches old content passively: memorized art history, generic literature appreciation, or detached theory.
The safe-bet version would include:
Creative production — writing, visual arts, film, music, design, performance, digital storytelling.
Critical interpretation — rhetoric, media literacy, propaganda analysis, fallacy detection, narrative framing.
Cultural preservation — local history, Philippine culture, community memory, language, heritage.
AI-era authorship — human-AI co-creation, provenance, attribution, copyright, synthetic media ethics.
Civic meaning — journalism, public communication, democracy, institutional trust, social repair.
Philosophy and ethics — reasoning, personhood, responsibility, technology governance, limits of optimization.
So the verdict:
Correct. Arts and humanities departments remain a safe strategic bet — provided they are redesigned as AI-era meaning, culture, creativity, and civic-coherence departments, not as nostalgic legacy units.
Use this url to view the conversation.
Backgrounder here:
Applying Gemini 3.1 Pro lens here.
Applying Perplexity AI/Claude Opus 4.6 lens here.
Applying Grok 4.3 lens here.
Comments