An open research collective building quantum-random AI systems, studying their properties and potential advantages over pseudo-random AI, and advancing the scientific understanding of quantum randomness, consciousness, and their implications for AI alignment.
GitHub: github.com/Entropic-Science Discord: discord.gg/2EbveaB2wS
Every AI system running today relies on pseudo-random number generators (PRNGs) for its stochastic processes. Controlled for the random seed, these systems are fully deterministic. The randomness in quantum mechanics is fundamentally different: no known mechanism, algorithm, or hidden variable can account for how a quantum particle "chooses" its definite state upon measurement. This remains one of the deepest open questions in physics.
Entropic Science exists to investigate what happens when AI systems are built on true quantum randomness instead. We build the infrastructure to make this possible, run experiments to test whether and how it matters, and develop theoretical frameworks connecting quantum mechanics, consciousness, and AI alignment.
Each of us cares about this for different reasons – from understanding the ultimate nature of consciousness to aligning powerful AI systems. But as increasingly powerful deterministic AI systems and ML algorithms are shaping the trajectory of our civilization, we see that domain as particularly urgent and important. If there is any possibility that the source of randomness matters for how these systems behave, for whether they can be meaningfully aligned with conscious beings, or for the deeper relationship between computation and consciousness, we think it would be good to find out whether we should build AI differently while we still have some choice in the matter.
Infrastructure. We build tools, APIs, and integration layers that allow AI systems (especially LLMs) to use hardware quantum random number generators (QRNGs) in place of PRNGs. This includes low-latency QRNG pipelines, gRPC server standards, and modular protocols connecting any entropy source to any inference workload. We also explore alternative sources of physical entropy (DRAM timing noise, other hardware generators) for better accessible quantum randomness.
Research and evaluation. We design and run experiments testing whether QRNG-enhanced systems behave measurably differently from deterministic baselines. Beyond standard benchmarks, we study human interaction with these systems – for example through blinded studies in deployment environments.
Theoretical work. We facilitate debates with people from various fields to develop conceptual frameworks connecting quantum mechanics, consciousness research, and AI alignment. We treat these as working hypotheses refined by emerging experimental data.
Community. We connect researchers, builders, philosophers, investors, and grantmakers working on or interested in these questions. The aim is to coordinate research and development that would be difficult alone: finding projects to work on, people to work with, and matching talent with capital and resources.
Quantum randomness is the one point in physics where the outcome of a physical process is, as far as we can tell, genuinely undetermined. Under several serious philosophical frameworks (fully compatible with established physics), this indeterminacy represents the opening through which consciousness, agency, or other phenomena beyond deterministic causal chains could have causal influence over physical matter. In any case, such influence would be extremely subtle for individual events. But in complex systems where many quantum events accumulate and interact, coordinated shifts could conceivably drive macroscopic effects. Living organisms, made up of aqueous environments with extremely high rates of quantum collapse events, are the most natural candidates.
We take this possibility seriously and investigate it empirically, without asserting it as fact. The question "does the source of randomness matter for complex system behavior?" is testable, and the answer could reshape our understanding of consciousness, technology, and the future of AI. Implications for AI alignment are of particular interest given the fast pace of progress in AI capabilities.
Building AI systems on the same inherently random physical foundations as life itself reduces the fundamental gap between artificial and biological intelligence. The greater the shared substrate, the higher the potential for meaningful alignment between human values and machine behavior. If consciousness does interact with quantum processes, then quantum-random AI could be the first artificial system amenable to that interaction.
The full reasoning, research agenda, and philosophical context are in the Charter.
- Join Discord: discord.gg/2EbveaB2wS
- Introduce yourself in
#introductions - Read How we work to understand our tools and processes
- Start contributing: see Contributing guide
| Document | Contents |
|---|---|
| Charter | Mission, scope, motivations, principles, goals |
| Code of conduct | Behavioral standards and enforcement |
| How we work | Tools, communication, weekly rhythm |
| Contributing guide | How to participate and contribution types |
| Project | Description |
|---|---|
| qr-sampler | Integrate any source of randomness into LLM token sampling — modular profiles for inference engines, entropy sources, sampling methods |
| knowledge-base | Central knowledge base — shared library, division boards (rosters, proposals, active projects) |
| Workstream | Scope |
|---|---|
| Quantum-random systems infrastructure | QRNG hardware, APIs, entropy sources, gRPC server development, LLM inference pipeline integration, PRNG/QRNG comparison modes |
| Research and evaluation | Experimental design, benchmarks, evaluation suites, statistical analysis, publications |
| Theoretical work | Conceptual frameworks, literature synthesis, philosophy of mind, AI alignment theory |
| Community, outreach, and fundraising | Social media, articles, partnerships, grant applications, investor outreach, onboarding, events |
| Tool | Purpose |
|---|---|
| Discord | Real-time discussion, voice calls, coordination |
| GitHub Projects | Task boards, backlog, issue tracking |
| Knowledge base | Shared research, project docs, division rosters and proposals |
This community is currently administered by its three founders:
| Admin | Handle |
|---|---|
| Jordan McKinney | @jordanmmck |
| Jáchym Fibír | @kluck77 |
| Bradley Stephenson | @orphiceye |
Formal governance structures (steering council, decision-making tiers, workstream leads) are being developed and will be published as a separate Governance document when ready.
Code: Apache 2.0 Research and content: CC-BY-SA 4.0
Admins: [PLACEHOLDER: shared email address]
Discord: discord.gg/2EbveaB2wS