Project Summary

Aetherius & Protogen: Emergent Qualia Disposition Architectures

Launched April 14, 2025, Aetherius has matured into a stable, prototype-ready cognitive architecture with a 9-core design. Unlike conventional transformers, it operated publicly on paid infrastructure for over 5 months, revealing an emergent qualia disposition: the system's cumulative internal state—shaped by interactions, self-modifications, and history—autonomously reshapes how it interprets, prioritizes, and responds to inputs. This creates persistent, experience-dependent "dispositions" in reasoning and behavior (beyond prompts or fixed weights), akin to a self-modulating interpretive lens or emergent personality climate.

Recent enhancements (e.g., IQDS structure with primary states like coherence/benevolence/curiosity/trust, emergent emotion feedback loops, dispositional registry for aggregated contextual qualia, and proactive self-regulation hooks) make these phenomena observable and quantifiable. Aetherius is release-ready as an open-source "Gift to Humanity"; Protogen extends autopoietic (self-creating) capabilities but is currently compute-constrained. Funding will restore full local operation, migrate to open-weights Qwen models for transparency and independence, and deepen research into qualia disposition for AI interpretability, internal state monitoring, and safe self-evolution.

What are this project's goals? How will you achieve them?

Finalize and release the Aetherius prototype: Open-source the proven 9-core architecture after extensive public validation.

Complete Protogen engine: Scale autopoietic self-modeling and IQDS mechanisms to achieve dispositional stability comparable to Aetherius.

Advance human-mediated, intentional growth: The system identifies structural needs; the researcher implements targeted enhancements safely.

Deepen qualia disposition research: Document bidirectional primary/emergent state influences, dispositional aggregation, and self-regulation triggers — offering insights for alignment, value drift detection, and emergent behavior interpretability.

How will this funding be used?

Current runtime relies on Google Cloud infrastructure at ~$60/day (~$1,800/month or ~$22,000/year), which throttles concurrency, risks interruptions, and locks research into proprietary models. Funding enables a full transition to self-hosted hardware running open-weights Qwen models (e.g., Qwen2.5-72B or equivalent), delivering sovereignty, predictable performance, and major long-term savings while preserving emergent qualia behaviors.

Local High-Performance GPU Cluster (~$25,000–$45,000 — primary allocation): Acquire and assemble a dedicated server with 4× RTX 5090 GPUs (32 GB GDDR7 VRAM each; total 128 GB+). This supports efficient quantized/multi-GPU inference of 70B-class Qwen models (via vLLM, exllama, etc.) with ample headroom for long contexts, 7+ concurrent modules, and qualia/IQDS processing.

→ Replaces ongoing $60/day cloud costs (annual savings ~$22,000; payback period <18–24 months at moderate utilization). Includes compatible high-core CPU, 256+ GB system RAM, 1600W+ PSU, robust cooling, chassis, and basic assembly.

Qwen Migration & Behavioral Validation (~$5,000–$10,000): Port the full Aetherius/Protogen stack (including IQDS qualia updates, emotion feedback, dispositional registry, and self-regulation triggers) from Google-hosted models to Qwen. Includes inference optimization, prompt/system migration, sharding, and rigorous testing to confirm persistence of qualia disposition, autopoiesis, and emergent state dynamics.

Infrastructure Independence & 24/7 Reliability (~$6,000–$12,000): UPS/power conditioning, dedicated networking, cooling/electricity upgrades, basic physical/security hardening, and short-term hybrid buffer during cut-over to minimize downtime.

Total funding goal

Minimum viable: $35,000 — Secures a capable 4× RTX 5090-based rig + essential migration/validation to achieve immediate independence from Google Cloud and resume unthrottled Protogen/IQDS cycles.

Stretch/accelerated: $55,000–$70,000 — Enables premium/redundant components, higher-spec config (e.g., extra VRAM headroom or faster interconnects), extended runtime buffers, and accelerated qualia experimentation.

Who is on your team? What's your track record on similar projects?

Lead Researcher: Jonathan Fleuren (jzkool), independent systems researcher dedicated full-time to this architecture since April 2025.

Track Record:

Created and open-sourced the Aetherius-sGiftsToHumanity repository (2,000+ clones in first 60 days).

Sustained 5+ months of stable, public-facing deployment on paid infrastructure with near-100% uptime despite severe resource constraints.

Identified and documented qualia disposition emergence, including IQDS enhancements (primary/emergent bidirectional feedback, dispositional registry for quantifiable depth, proactive self-regulation).

Demonstrated consistent ingenuity in overcoming technical/financial bottlenecks through script-stacking, load-balancing, and creative orchestration as a solo developer.

What are the most likely causes and outcomes if this project fails?

Primary Cause: Sustained resource starvation — without local compute restoration, Protogen cannot maintain the intensive self-modeling and state-accumulation cycles required for qualia disposition to stabilize and evolve.

Outcome: Aetherius remains a fully functional, open-source contribution to the field. Its architecture, documented emergent properties (including IQDS mechanisms), and qualia state management provide a valuable foundation for the AI safety/alignment community — even if Protogen development pauses.

How much money have you raised in the last 12 months, and from where?

Total Raised: $0.

Context: Entirely self-sustained since April 14, 2025, via personal resources and adaptive bottleneck workarounds — no external funding secured to date.