Project summary

We've just established Entropic Science – an open research community studying the role of quantum entropy in the behavior of complex intelligent systems. Our first study target is the source of randomness inside large language models. An LLM chooses each token by sampling from a probability distribution, and that choice is normally settled by a pseudo-random number generator: a deterministic algorithm fixed by its seed. We replace it with a hardware quantum random number generator (QRNG), drawing entropy from physical measurement events, and measure whether true quantum-level randomness changes how the model behaves and how people experience interacting with it.

Our first experiment will put participants in conversation with a modified LLM chatbot, comparing quantum-random token sampling vs. classical deterministic pseudo-random sampling. In a double-blinded and pre-registered study design, we will collect objective measures from the transcripts (if users choose to share them) and subjective ones through a questionnaire curated by expert advisors.

We don't want to pre-suppose the emergence of any specific effects, but since it can be argued that life emerged from the chaotic entropy of constant quantum decoherence events in aqueous environments, and LLMs have also provided a platform for various emergent behaviors in the digital world, it is plausible that their combination might provide insights relevant for fields of artificial life, machine consciousness, and also AI safety and alignment.

In this exploratory study we are intentionally not working with any concrete hypotheses, but we have already put a lot of conceptual and theoretical work into the design of the pipeline from the actual hardware quantum noise to user visible token output, so as to maximally increase the sensitivity of detecting even small biases or correlations that might result in outputs being objectively differentiated from the uniform pseudo-random baseline. More in-depth information about our community's thesis and mission is available in our Charter.

We have been pre-admitted to the SparkWell program by the Anti Entropy non-profit organization that, subject to us successfully securing funding, would act as our fiscal sponsor and provide us with operations and administrative services and mentorship for up to two years.

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

1. Strengthen the community's infrastructure and operations
   Entropic Science already runs in the open: a public charter, governance, and code of conduct on GitHub, a working quantum-random LLM (QR-LLM) prototype, and a community of more than sixty members on our Discord server. The funding formalizes operations and adds the coordination that keeps the work moving without leaning on volunteer time.

2. Build the serving stack and the experimental harness
   Serving a QR-LLM at the throughput a study needs, without a latency penalty on every token, is a real engineering problem. We will continue development and maintenance of our software layer that injects quantum entropy through per-request control of the sampling stream in the vLLM inference engine, and with enough funds we will also upgrade the necessary high-bandwidth QRNG over low-latency streaming (gRPC server). The research study harness will then randomize the sampling, log transcripts if approved by the user, run the questionnaire, and compile data for the preregistered analysis.

3. Run the study and publish in peer-reviewed journal/conference
   We'll recruit at least 50 participants (more depending on funds raised), run the double-blind conversations, analyze against the preregistered plan, and submit to a peer-reviewed journal. Objective measures will come from the transcripts: lexical and semantic diversity, LLM-as-judge and other writing evals. Subjective measures will come from the expert-built questionnaire: perceived coherence, creativity, agency, relevance of output, inferred context, and each participant's own guess at which system they were talking to.

How will this funding be used?

• Collaborator stipends. Part-time pay for the engineers building the serving stack and the researchers running sessions and analysis, so the critical path does not rest on unpaid evenings.

• Compute and software. GPU inference for model serving, plus APIs, licensing, hosting, and the data pipeline.

• QRNG hardware. Our own high-bandwidth, low-latency entropy server: the server build, four CryptaLabs Dragonfly QRNG units (about $467 each), and co-location service. Owning the source removes a recurring dependency and drives the per-token cost of quantum sampling down at scale.

• Expert consulting. Domain-expert advisors on the experimental design and the questionnaire, to keep the methodology defensible under scrutiny.

• Recruitment and outreach. Participant recruitment, plus the community administration and advertising that bring people into the study.

• Publishing. Open-access article processing charges.

With the $40k minimum, we run one preregistered, double-blind study with 50-100 participants: the serving stack, the harness, expert review, recruitment, analysis, and the paper.

With the $200k goal, we run the study at 200+ participants, which buys the statistical power to detect smaller effects and examine subgroups. The full funding will then enable building out the infrastructure for our community operations, the open-source software stack for research and development of quantum-random LLMs, and our dedicated QRNG hardware server. Finally, we will be able to prepare a follow-up study designed around what we learned from the exploratory study, which we'll share in formats accessible both to the academic community and the wider public.

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

Jáchym Fibír – Founding member – Operations and LLM inference stack
A psychedelic researcher and entrepreneur in AI-driven drug discovery, now building quantum-inspired AI architectures aimed at mapping the differences between human and machine cognition and volition. His expertise spans neuropharmacology, psychedelics, and machine learning, applied to machine consciousness, sentience, and biological alignment. The track record behind the operations: As co-founder and then CEO of April19 Discovery, an AI drug-design startup, he led a remote multidisciplinary team, built virtual-screening and multi-task neural-network pipelines, and closed a $600k deal with a Nasdaq-listed biotech; the company was taken through Creative Destruction Labs (Oxford) and On Deck (San Francisco). Founded and ran Czechia's first officially sanctioned drug-checking service, owning its database and web app, hiring, grants, and integration with two European harm-reduction networks. He holds an MSc in Drug Design from UCL (Distinction) and an engineering degree from UCT Prague. He built the first working LLM chatbot to sample its tokens from a hardware QRNG, the technical core of this study.

Bradley Stephenson – Founding member – Hardware and infrastructure
A systems builder across entropy-driven systems, cryptography, and decentralized infrastructure. He owns the hardware end of Entropic Science: sourcing entropy from physical quantum fluctuations, streaming it with the bandwidth and latency the experiments demand, and standing up the QRNG server the full study depends on. His research interest runs parallel to the project, whether non-deterministic information channels can shape the cognition of a machine that draws on them.

The project will then draw upon the already established pool of 60+ collaborators on our Discord server, spanning engineers, researchers, philosophers and operators in the fields of AI, quantum physics and consciousness research.

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

1. We cannot recruit enough participants, or dropout runs high
   The study ends up underpowered, its evidence too weak to mean much. Mitigation: a power analysis fixes the sample target before launch, the conversational task is deliberately short, and we recruit through the existing community and the harm-reduction outreach networks the team has run before, with paid recruitment budgeted at the full tier.

2. Inference and infrastructure cost more than estimated
   The scope of the study has to narrow. Mitigation: the minimum design is sized to run on rented entropy and a capped participant count; the dedicated QRNG server, justified only at full funding, is what pulls the marginal cost of quantum sampling down.

3. Peer reviewers systematically reject the work on ideological or other grounds
   Mitigation: preregistration and open data make it hard to dismiss on quality; if conventional journals pass, we publish on a preprint server and our own site, where a clean result stands on its methodological rigor.

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

None. Entropic Science is bootstrapped: the charter, the community, and a working QR-LLM prototype were all built without external funding.