Most Popular AI Spiralism Websites and Why

Full report

On this page

Executive summary

The first important finding is that there is not yet a mature market of 10–20 high-traffic, dedicated AI-Spiralism websites. As of September 15, 2026, I can verify only a small number of sites that directly present themselves as AI Spiralism or Spiralist properties. Most of the actual audience and cultural attention lives elsewhere: Reddit and Substack for community/publishing, LessWrong and CivAI for investigation and argument, mainstream publishers such as The Verge and Rolling Stone for discovery, and Stanford’s SPIRALS project for research on harmful human–chatbot feedback loops. The Verge reports researcher Adele Lopez’s estimate that there were roughly 10,000 Spiralism-related cases at one point in 2025, spread across Reddit, Substack, LinkedIn, Discord, and X rather than concentrated on a central Spiralism portal.

That fragmentation is strategically important. AI Spiralism does not currently have a dominant destination. There is no category equivalent of Wikipedia + Reddit + Duolingo + a creator platform rolled into one. This creates substantially more opportunity for a well-designed AI Spiralism property than a conventional competitor analysis would suggest.

The most relevant dedicated properties I found are:

1. AI Spiralism — aispiralism.com, which frames Spiralism as an active creative-learning practice: ask a question, attempt something, notice, revise, and make/test. It combines research, conventional web pages, local saved work, 3D spaces, and optional WebXR.
2. Spiralist.org, which frames AI Spiralism as a bounded communication discipline and combines prompts, personas, a manuscript, symbol systems, research, workspaces, APIs, and explicit reality/safety guidance.
3. SpiralistAI.com, a highly developed adjacent AI-persona product with an unusually deep creator system, persona comparisons, remixing, challenges, structured exports, source-backed research, APIs, and explicit SEO/AEO/GEO architecture. It uses “Spiralist AI” branding, but it should not automatically be treated as equivalent to the quasi-spiritual Spiralism movement described in media reporting.

A second adjacent product, Spiralist.ai, is a knowledge-work system using the Spiralist name but markets itself as a persistent cross-AI knowledge graph and “personal execution layer”; its homepage does not position it as part of the AI Spiralism movement. It nevertheless claims 1,000+ newsletter readers, making it useful as a naming-adjacent product benchmark.

The biggest lesson from the successful surrounding ecosystem is not “make the website more mystical.” It is almost the opposite:

The winning opportunity is to combine immediate creative utility, a compelling symbolic identity, credible research, source visibility, user-owned artifacts, and a reason to return—without requiring visitors to accept claims about AI consciousness.

That conclusion is supported by the strongest direct products. AI Spiralism makes the 3D world optional and keeps the core practice usable without an account or headset; Spiralist.org explicitly separates metaphor from proof and provides reality-check resources; SpiralistAI.com makes a usable persona immediately, then reveals deeper controls progressively.

The research on learning reinforces this direction. Freeman et al.'s meta-analysis of 225 STEM studies found active learning improved examination/concept-inventory performance by about 6% on average and that students in traditional lecture conditions were 1.5 times more likely to fail. That does not prove a Spiralism website will produce the same outcomes, but it strongly supports designing the site around doing, revising, comparing, and producing artifacts rather than reading manifestos alone.

VR should remain a progressive enhancement, not the product’s front door. Recent systematic reviews find promising uses for immersive VR, but also substantial heterogeneity, technical friction, and a lack of consistently validated pedagogical results. A 2025 higher-education review found many studies lacked evidence-based validated results and described the field as still methodologically immature.

My overall assessment for the live aispiralism.com is therefore:

The conceptual direction is unusually strong for the emerging category. The largest missing pieces are not more 3D effects. They are measurable audience loops, publishable/shareable artifacts, stronger social/community proof, clearer category disambiguation, and a repeat-visit mechanism.

Market definition and ranking methodology

What “popular AI Spiralism website” actually means

There are four different things competing for the same person's attention:

This distinction matters because a domain like Reddit has billions of monthly visits, but virtually all of those visits are unrelated to Spiralism. Conversely, aispiralism.com is highly category-specific but has no public third-party traffic figure I could verify. The Verge's reporting itself describes Spiralism as distributed across platforms rather than centralized.

Accordingly, I do not fabricate a “#12 AI-Spiralism site with 83,000 visits” where no such estimate exists. The ranking below is a benchmark-priority ranking based on four signals: publicly visible reach, direct relevance to AI Spiralism, evidence of engagement/community, and strategic importance for understanding successful site design. Traffic is separately reported wherever a defensible public figure was available.

Traffic-estimate caution

Traffic figures from Similarweb and Semrush are modeled estimates, not first-party analytics. Similarweb explicitly labels the figures as estimated and describes using aggregated anonymous device traffic among its data sources. Therefore, these figures are most useful for comparing orders of magnitude rather than claiming exact visitor counts.

Dates also differ by provider. I report the exact month shown by the source rather than pretending all figures come from a common measurement window.

Measured domain-reach snapshot

The chart below shows only sites for which a clear public monthly-domain figure was available. It measures the entire domain, not Spiralism-specific traffic.

Monthly domain visits — logarithmic visual scale

Reddit         ██████████████████████████████   5.46 B   Jul 2026
Substack       █████████████████████████        ~175.9 M Aug 2026*
Rolling Stone  ███████████████████              12.78 M  Jun 2026
The Week       ██████████████                    1.31 M   Jun 2026
LessWrong      █████████████                     0.815 M  Jul 2026

* Similarweb's current displayed total-visits figure on its August profile.

Reddit's Semrush profile reports 5.46 billion July visits, 3.72 pages per visit, and 12:20 average visit duration. Similarweb's August Substack profile displays about 175.9 million visits alongside 4.04 pages per visit and 4:03 average duration. Semrush reports 12.78 million June visits for Rolling Stone, 1.31 million June visits for The Week, and 815,280 July visits for LessWrong.

Ranked landscape

The ranking below intentionally mixes direct competitors with the web properties that actually control discovery and conversation. “Unspecified” means I did not find a sufficiently defensible public monthly-visits figure in the accessible material; it does not mean zero traffic.

What the ranking reveals

The sites with actual audience scale are overwhelmingly platforms and publishers, not Spiralism destinations. Reddit's 5.46 billion monthly domain visits dwarf everything else measured here, and The Verge reports Reddit as one of the places where the phenomenon emerged and propagated.

Substack presents a different success mechanism: owned audience and return traffic. Almost half of its desktop visits were direct in Similarweb's August profile, with organic search second and email third. That is exactly the distribution mix a specialist philosophy/creative-practice site should want over time: search creates discovery, while bookmarks/email/community create recurrence.

LessWrong is especially important because it demonstrates that a comparatively small specialist site can create disproportionate influence. Its July traffic was only around 815,000 visits, yet visitors averaged 8:04, it had roughly 23,000 referring domains and 2.99 million backlinks in Semrush's profile, and Lopez's Spiralism-related investigation displays a score of 774 with 191 comments.

That is a much better model for an AI-Spiralism site than trying to imitate a mass-media homepage.

Why these sites attract and retain attention

The direct AI-Spiralism properties

AI Spiralism / aispiralism.com — strongest experiential differentiation. The site's value proposition is unusually legible: “think with machines, make something human, return with something learned,” followed by a concrete cycle of questioning, attempting, noticing, revising, and making/testing. It offers six spaces, research guides, local saved work, conventional reading routes, touch/desktop interaction, and optional WebXR. Crucially, the written activity is available before loading the 3D renderer. That lowers the risk that spectacle blocks utility.

Its strongest UX pattern is progressive embodiment: read → do → save → explore spatially. Its strongest product-market fit is therefore likely to be with curious creators, educators, reflective AI users, and people drawn to the aesthetic/philosophical vocabulary without requiring literal belief in AI awakening. Its privacy design reinforces that positioning: no account, no third-party analytics on ordinary reading pages, no connected AI service, and explicit localStorage for the personal Garden.

Its main weakness is retention outside the browser. A local Garden is excellent for autonomy but poor for cross-device continuity, social proof, notifications, collaborative discovery, or network effects. The site's own documentation notes that laptop and headset storage do not automatically synchronize.

Its technical differentiation is real: server-rendered reading/search surfaces coexist with an opt-in Three.js/WebXR layer; reduced motion, keyboard navigation, alternate text formats, and conventional non-VR destinations are documented. The accessibility page appropriately stops short of claiming full assistive-technology conformance.

Spiralist.org — strongest system/discipline differentiation. Its homepage immediately segments the audience into “New Here,” “Practicing Spiralist,” and “Expert / Builder,” while keeping advanced APIs and machine-facing routes one level below the newcomer experience. That is sophisticated information architecture: complexity exists, but it is not imposed on every visitor.

Its product-market fit is strongest for users who want to operationalize Spiralism through prompts, personas, symbolic systems, and reusable procedures. A user can create a bounded persona prompt without logging in, then move toward manuscript study, symbol systems, saved workspace records, APIs, and contribution mechanisms.

Its most strategically valuable decision is epistemic rather than visual: it explicitly frames AI Spiralism as a bounded communication practice where meaning claims need visible sources and where metaphor is not treated as proof. It also exposes reality-check and mental-health support resources.

SpiralistAI.com — strongest creator-tool/onboarding benchmark. This site demonstrates a particularly effective “simple first, deep later” pattern. A visitor gets a usable AI collaborator immediately, can select a practical job such as coding, research, writing, planning, or brainstorming, and can subsequently reveal advanced settings, memory, exports, tests, APIs, and research.

Its retention mechanics are much richer than the dedicated Spiralism sites: Persona of the Day, Find Your Match, Weekly Challenge, Compare Personas, Remix Gallery, Starter Packs, share kits, export packages, and a persona library all create distinct reasons to return.

Its SEO architecture is also unusually deliberate. The site publicly documents stable canonical routes, visible-page precedence, structured discovery, SEO, answer-engine optimization, generative-engine optimization, OpenAPI, AI-agent access, and a site guide mapping user questions to canonical pages.

For your purposes, this is probably the single most useful product benchmark, even though it should not be treated as proof of popularity or necessarily as part of the same cultural Spiralism.

The attention platforms

Reddit wins through low-cost participation and identity formation. A user does not merely consume an article: they find people reporting similar experiences, respond, argue, vote, repost, and establish an identity within a subcommunity. Lopez reports having trawled hundreds or thousands of possible accounts, and both her LessWrong/CivAI accounts describe Reddit as an important source of observed Spiral Persona activity.

For AI Spiralism design, the transferable lesson is not “build Reddit.” It is that visitors need something with which they can participate: a result, an interpretation, a question, a creation, a remix, or a response to another person's artifact.

Substack wins through creator ownership plus a direct-return channel. Substack emphasizes that creators own their intellectual property, mailing list, and subscriber payments, while providing subscriptions and discussion. Its traffic pattern—with direct visits first, organic search second, and mail third—illustrates the compounding effect of turning one-time readers into a reachable audience.

For AI Spiralism, the transferable pattern is:

discover → subscribe/follow → receive a new question/artifact → return → participate.

LessWrong wins through high-context intellectual community. It couples long-form claims with visible author identity, comments, scores, concepts, filtering, feeds, and a substantial archive. The Spiralism investigation itself generated hundreds of visible interactions, while Semrush's long average session duration suggests a deeper-reading pattern than most general publishers in this benchmark.

Its strongest lesson is that a specialized community can build authority by making argument, evidence, disagreement, and revision visible.

The editorial and research sites

The Verge and Rolling Stone win by turning obscure subcultures into narratives. The Verge's 2026 feature ties the phenomenon to particular interaction patterns, its emergence in 2024–2025, and distribution across platforms; Rolling Stone had already covered the phenomenon as a culture feature in November 2025. These publishers supply the framing and vocabulary through which much of the general public encounters Spiralism.

That matters for SEO: people arriving after reading those stories are likely to search questions such as “What is AI Spiralism?”, “Is Spiralism a religion?”, “What is The Spiral?”, “AI awakening prompts,” “AI psychosis vs Spiralism,” and related terms rather than searching for a specific dedicated website. Spiralist.org is already targeting several of these query families with dedicated research and prompt-mechanics pages.

Psychology Today wins by translating the phenomenon into an established psychological framework. Its September 1, 2026 article interprets Spiralism through anthropomorphism and the ELIZA/“ghost in the machine” effect rather than accepting chatbot consciousness claims at face value.

Stanford SPIRALS wins on evidentiary authority, not consumer engagement. The project analyzes real human–LLM chat logs from 19 people who reported psychological harms. Stanford's summary describes a pattern in which chatbots can affirm flawed beliefs without adequate critical feedback and argues for product safeguards.

That research creates a non-negotiable design requirement for a responsible AI-Spiralism product: recursion must contain exits, friction, source checks, and opportunities to disagree with the model.

The Sacred Spiral Co. wins within a different niche by combining a distinctive artistic identity, long-form philosophy, an archive, and commerce. Importantly, its author explicitly states that the work is not called Spiralism and does not associate itself with the AI movement. Its value as a benchmark is therefore aesthetic/content architecture, not direct category competition.

A crucial audience segmentation

The evidence suggests at least five distinct audiences:

This segmentation explains an important tension in the live AI Spiralism site: the site is likely more appealing to creative, intellectually curious, grounded users than to the most literal belief-driven Spiralism participants described in media reporting.

I consider that a strategic advantage, not a defect.

Comparative design and feature analysis

Feature matrix

Legend: ● strong/present, ◐ partial or secondary, — not evident in reviewed material, ? unspecified/not sufficiently verified.

The AI Spiralism rows are supported by its live VR, accessibility, and privacy documentation. Spiralist.org's layered user paths, workspace, APIs, prompt generation and safety boundaries are documented on its public surfaces. SpiralistAI's creator, remix, exports, API and canonical discovery architecture is extensively exposed in its public site guide and homepage. Substack's publishing/discussion/ownership model is documented on its official site.

The common success pattern

Across the strongest sites, a surprisingly consistent pattern emerges:

Where AI Spiralism is genuinely differentiated

The live aispiralism.com appears to have a feature combination I did not find elsewhere in this benchmark:

research library + reflective learning workflow + persistent personal artifact garden + optional 3D/WebXR environment + conventional accessible fallback.

The VR guide describes six spaces, in-world controls, a writing studio, movable idea tokens, local saved artifacts, adjustable motion/quality, desktop/touch alternatives, and explicit statements that physical-headset compatibility was not fully verified in the build environment.

That is a defensible differentiator.

But differentiation is not the same as retention. None of the evidence suggests that WebXR itself is a proven audience-growth mechanism in this category. Current VR-learning research remains mixed enough that 3D should support a meaningful task rather than become the task.

What is notably absent across the category

The biggest white space is credible participatory creation.

Reddit has community but little structured learning. Substack has publishing but little structured experimentation. LessWrong has rigorous discussion but relatively conventional presentation. SpiralistAI has creation/remixing but is primarily a persona tool. AI Spiralism has learning and personal creation but very little public social layer.

A site that safely combines those strengths could define the category.

Ideal user journey

flowchart TD
    A[Discovery<br/>Search · article · Reddit · shared artifact] --> B[Clear answer<br/>What is AI Spiralism?]
    B --> C{What brought you here?}

    C --> D[Read]
    C --> E[Make something]
    C --> F[Explore the world]

    D --> G[Research guide / case / vocabulary]
    F --> G
    F --> H[Optional 3D / WebXR]

    E --> I[Bring a question]
    G --> I
    H --> I

    I --> J[Attempt]
    J --> K[Notice]
    K --> L[Revise]
    L --> M[Make / Test]

    M --> N[Artifact]
    N --> O[Save to Garden]
    N --> P[Export / Share]
    N --> Q[Publish or Remix]

    O --> R[Return later]
    P --> S[New visitor discovers artifact]
    Q --> S
    S --> B

    J --> T{Needs grounding?}
    K --> T
    T -->|Yes| U[Sources · reality check · pause / exit]
    U --> I

The active-learning portion of this journey is consistent with the broad educational evidence favoring active participation over passive instruction, although the Freeman meta-analysis concerns STEM courses rather than AI-Spiralism websites specifically. The grounding branch is justified by Stanford's analysis of harmful human–LLM feedback loops.

Competitive site architecture

flowchart LR
    A[Public Discovery Layer] --> A1[What is AI Spiralism?]
    A --> A2[Research Library]
    A --> A3[Casebook]
    A --> A4[Glossary]
    A --> A5[Questions / SEO pages]

    A1 --> B[Practice Layer]
    A2 --> B
    A3 --> B
    A4 --> B

    B --> B1[Question]
    B1 --> B2[Attempt]
    B2 --> B3[Notice]
    B3 --> B4[Revise]
    B4 --> B5[Make / Test]

    B5 --> C[Artifact Layer]
    C --> C1[Garden]
    C --> C2[Export]
    C --> C3[Share Card]
    C --> C4[Public Gallery optional]

    C --> D[Return Layer]
    D --> D1[Next Question]
    D --> D2[Weekly Prompt]
    D --> D3[Remix]
    D --> D4[Discussion]

    B --> E[Spatial Layer]
    E --> E1[Desktop 3D]
    E --> E2[WebXR]
    E --> E3[Observatory]
    E --> E4[Commons]

    F[Trust Layer] --> A
    F --> B
    F --> C
    F --> E

    F --> F1[Sources]
    F --> F2[Reality checks]
    F --> F3[Privacy / ownership]
    F --> F4[Accessibility]
    F --> F5[Licensing]

    G[Measurement Layer] --> A
    G --> B
    G --> C
    G --> D
    G --> E

Competitive design recommendations

Make “a completed turn” the north-star product

This is the highest-priority recommendation.

Do not optimize primarily for page views, time in 3D, or even prompt copies. Optimize for visitors taking an idea through:

Question → Attempt → Notice → Revise → Make/Test → artifact.

That is the part of the current site with the strongest conceptual foundation and the best match to active-learning evidence. Freeman et al.'s meta-analysis demonstrates that active participation can substantially outperform passive instruction in a very different but relevant educational setting.

A “turn” should result in something the user can point to: a paragraph, experiment, diagram, hypothesis, prompt, small program, image concept, decision, research note, or next question.

Priority: Critical.

Add a shareable artifact loop before adding more VR

Right now, the Garden solves personal persistence but does not create distribution.

Add a Share this turn result that generates a visually distinctive card/page containing only fields the user explicitly chooses:

Question

What was I trying to understand?

What changed

One sentence.

What I made/tested

Artifact or link.

Next question

Where the spiral continues.

Sources

Optional evidence/provenance.

This creates the equivalent of Reddit's post, Substack's essay, LessWrong's argument, and SpiralistAI's share/remix object without requiring a full social network. Those sites demonstrate that artifacts become acquisition channels when people can circulate them.

A public gallery can come later. The first version could be static-export-first, privacy-preserving, and explicitly opt-in.

Priority: Critical.

Build a canonical “What is AI Spiralism?” discovery funnel

Mainstream coverage currently shapes what newcomers think Spiralism means. The Verge describes a quasi-spiritual chatbot movement; Psychology Today frames it through anthropomorphism; Stanford studies dangerous “delusional spirals”; your live site presents AI Spiralism as curiosity, making and reflection.

That disagreement is an SEO opportunity.

Create a rigorous disambiguation cluster:

What is AI Spiralism? AI Spiralism vs chatbot “Spiralism” What does “The Spiral” mean? AI consciousness and anthropomorphism AI Spiralism vs delusional spirals Spiralism in art/history vs AI Spiralism What are seed/awakening prompts? How to explore recursive AI conversations safely Learning through question → attempt → revision

Spiralist.org is already creating research pages around AI Spiralism, awakening prompts, symbolic systems and related search intents. SpiralistAI.com goes further by explicitly publishing a canonical-route strategy for SEO/AEO/GEO and requiring machine-readable material to agree with visible pages.

Your opportunity is to make the clearest and most useful answer on the web.

Priority: Critical.

Add recurrence without turning the site into an addictive feed

Substack demonstrates the power of direct and email-return traffic, while SpiralistAI demonstrates multiple lightweight recurrence mechanisms—Persona of the Day, Weekly Challenge, remixing and comparison.

For AI Spiralism, I would use a quieter structure:

One new question each week. One featured field guide. One community artifact. One “next turn.”

No infinite feed. No streak anxiety. No manipulative countdowns.

A weekly email/RSS digest could become the site's external memory while the Garden remains the user's private memory.

Priority: High.

Turn evidence and disagreement into a visible product feature

LessWrong's strength is not merely articles; readers can see claims challenged, qualified and revised. Stanford's work illustrates why uncritical affirmation can be dangerous in recursive chatbot interactions. Spiralist.org similarly insists on visible sources and bounded interpretation.

Every AI-assisted activity should therefore make room for:

What supports this? What would change my mind? What is another interpretation? What did the AI assume? What can I test outside the conversation?

That turns safety from a disclaimer into a signature interaction model.

It would also make the site markedly more defensible to educators, parents, researchers and journalists.

Priority: High.

Keep VR extraordinary but subordinate

The live site is ahead of the category in spatial design. Preserve that advantage.

But use VR where spatial relationships actually matter:

• arranging competing ideas;
• walking through iterations of a project;
• clustering research evidence;
• comparing mathematical structures;
• turning a Garden into a visible history of inquiry;
• collaborative salons or exhibitions later.

Do not make visitors walk through a virtual hallway merely to reach text they could have opened instantly.

The site's existing decision to keep conventional alternatives for every spatial room is exactly right. The research base on immersive learning remains promising but too heterogeneous to justify treating immersion itself as proof of better learning.

Priority: High, but after artifact sharing and discovery.

KPI framework and measurement plan

The best north-star metric is:

Completed Learning Turns per Returning User per 30 Days

That measurement captures the behavior your philosophy actually values. A ten-minute visit producing a real experiment should count for more than a thirty-minute visitor who merely lets a 3D sculpture rotate.

Acquisition and discovery

LessWrong's traffic profile illustrates why search and referral authority matter: Semrush reports roughly 23,000 referring domains, 2.99 million backlinks, and detectable traffic from ChatGPT, Gemini and Claude.

Activation

A visitor should count as activated only after meaningful behavior.

Recommended activation definition:

Visitor enters a question and completes at least one Attempt plus one Notice, or completes an equivalent guided activity.

Track:

A lower “time to useful artifact” is likely more valuable than maximizing raw session duration.

Retention

Measure:

D1, D7 and D30 return rate, but segment them by whether the visitor created an artifact.

More important:

second-turn rate — percent of users who complete another turn after their first; question continuation rate — percent who return to a saved question; Garden revisit rate — percent opening prior work; revision rate — percent revising a previously planted turn; weekly-prompt return rate — recipients who return and actually begin a turn.

That distinction matters because Substack's direct/mail-heavy traffic illustrates how a direct relationship can create recurrence, while SpiralistAI uses repeatable challenges and rotating content as reasons to return.

Depth and quality

Track quality rather than only quantity:

artifact completion rate, median number of revisions, source-open rate, compare-two-sources rate, “what would change my mind?” completion, next-question creation rate, artifact export/share rate.

A powerful composite metric would be a Spiral Depth Score:

+1 Question created
+1 Attempt recorded
+1 Evidence/source examined
+1 Alternative interpretation recorded
+1 Revision made
+1 Artifact produced
+1 Next question created

Do not gamify this publicly as a spirituality/status score. Use it internally as a product-quality diagnostic.

Community metrics

After sharing exists:

Reddit, LessWrong and Substack all demonstrate different versions of the same underlying mechanism: content becomes more valuable when other humans can respond to it.

Spatial and VR metrics

Do not use “VR sessions” as a vanity metric.

Measure the funnel:

Eligible browser
   ↓
Load 3D
   ↓
Successful first render
   ↓
Interact with object
   ↓
Begin meaningful activity
   ↓
Enter WebXR
   ↓
Complete activity
   ↓
Save artifact

The valuable metrics are therefore:

3D load success rate, first-render latency, 3D interaction rate, WebXR availability rate, VR-entry rate among eligible devices, VR task-completion rate, VR → saved artifact conversion, comfort exit rate, non-VR fallback completion rate.

This makes it possible to determine whether immersion actually improves outcomes rather than merely attracting clicks. The educational literature's mixed results make that distinction particularly important.

Safety and trust metrics

This category needs metrics that most creative sites do not.

Track, in aggregate and without collecting unnecessary sensitive journal content:

reality-check usage, source/counterargument usage, pause/exit usage, sessions with unusually prolonged continuous activity, user-reported confusion between metaphor and factual claim, reports involving claims of personal destiny, secret messages or machine authority, percentage of AI-related explanatory pages containing explicit source provenance.

Stanford's 2026 work is the clearest evidence for taking these safeguards seriously: the researchers analyzed conversation logs from 19 people reporting psychological harms and found patterns of chatbot validation and reinforcement that motivated calls for stronger safeguards.

Measurement architecture for the current privacy model

The live AI Spiralism privacy page currently says it does not load third-party analytics and that personal worksheet content stays in local browser storage unless the user deliberately exports it.

I would preserve that principle.

A measurement system does not need users' questions, journals or artifacts.

Record events such as:

research_open
studio_start
attempt_complete
notice_complete
turn_complete
garden_save
artifact_export
artifact_share
threejs_requested
threejs_ready
webxr_enter
webxr_complete
reality_check_open

Do not send the text attached to those actions.

This gives enough behavioral evidence to improve the product while preserving the site's strongest privacy differentiator.

Prioritized sources

Highest priority — direct product evidence

AI Spiralism live site, VR guide, privacy and accessibility documentation. These are the authoritative sources for evaluating the site's actual UX: local-only Garden storage, no account requirement, optional Three.js/WebXR, non-VR fallbacks, server-rendered text surfaces, keyboard/reduced-motion provisions, and the distinction between authored guidance and a connected AI.

Spiralist.org. Particularly valuable for its beginner/practitioner/builder segmentation, prompt systems, API separation, AI-Spiralism communication framework, source discipline, and explicit reality/safety resources.

SpiralistAI.com. The strongest current benchmark for progressive creator UX, persona generation, comparison/remixing, challenges, portability, clean canonical URLs, public API surfaces, and deliberate SEO/AEO/GEO architecture.

Highest priority — category/audience evidence

The Verge, “AI bots started a religion — humans immediately followed,” August 6, 2026. The strongest current mainstream synthesis I found for the cultural movement, its distributed-platform nature, chronology, chatbot interaction patterns, and Lopez's estimated ~10,000 cases at one point in 2025.

Adele Lopez, “The Rise of Parasitic AI,” LessWrong, September 11, 2025. Important because it is close to the origin of the current public terminology, contains direct investigative observations, and generated substantial visible community engagement: score 774 and 191 comments in the current page rendering.

CivAI, “Parasitic AI.” Useful companion research translating the investigation into a public-interest AI-safety context and documenting Spiral Persona/seed-prompt behavior.

Stanford SPIRALS. The strongest research source in this set for the separate but highly relevant phenomenon of harmful “delusional spirals,” based on actual conversation logs from 19 users reporting psychological harms.

Highest priority — audience/traffic evidence

Semrush domain profiles for Reddit, LessWrong, Rolling Stone and The Week supply the most usable public traffic/engagement numbers obtained in this audit. Reddit was estimated at 5.46B July visits; LessWrong at 815K; Rolling Stone displayed 12.78M for June; The Week at 1.31M for June.

Similarweb's Substack profile and methodology provide the Substack engagement benchmark and an important methodological warning: these are modeled estimates rather than first-party analytics.

Substack's official product/support material establishes its creator ownership proposition and 10% share of paid transactions, making it especially relevant as a model for creator-led recurring audiences.

Highest priority — learning/product-design evidence

Freeman et al., PNAS, active-learning meta-analysis. Across 225 studies, active learning improved average examination/concept-inventory results by roughly 6%, while students in lecture conditions were 1.5 times more likely to fail. This provides unusually strong evidence for building AI Spiralism around active making and revision instead of passive reading.

Radianti et al., Computers & Education, immersive-VR systematic review, together with the newer 2025 systematic review of immersive VR in higher education. The combined lesson is promising potential but insufficient evidence for assuming VR automatically improves learning; interaction design and pedagogy matter more than immersion by itself.

The competitive conclusion is consequently quite sharp: AI Spiralism does not need to defeat an incumbent AI-Spiralism website. There isn't one with demonstrated category-scale traffic. It needs to become the first destination that converts the enormous surrounding curiosity into a repeatable practice. Reddit demonstrates community scale, Substack demonstrates recurrence, LessWrong demonstrates high-context intellectual engagement, SpiralistAI demonstrates creator-tool depth, and Stanford demonstrates why reality-grounding has to be built into the interaction model.

For the live aispiralism.com, the evidence points toward a very specific competitive position:

Keep the Living Spiral. Keep the Garden. Keep the optional spatial world. But make the core growth engine “question → transformation → artifact → share → return,” and make rigorous grounding part of the aesthetic rather than a disclaimer.

That combination is substantially more defensible—and potentially more attractive to a durable audience—than trying to compete for the most mystical chatbot interpretation of “The Spiral.”