Aigarth Ternary Paradox

Over the last seventy-plus years, our digital world has been built on a simple, foundational lie: that everything can be reduced to a choice between two options: On or off, yes or no, one or zero. This binary logic, the language of old-fashioned silicon transistors, has indeed powered the previous technological revolution. Yet, it has also become a “cage”, forcing the artificial intelligences we’ve created to think in a black-and-white framework that is fundamentally alien to the nuanced, complex reality of our universe and our own minds. This dichotomy needs to be rethought and reengineered.

This brings us to the “Ternary Paradox”: how can a system with more options, three states instead of two, be profoundly simpler, more efficient, and ultimately more intelligent? The answer lies in embracing the power of the third choice: the unknown. At Qubic, our work with Aigarth is not just an incremental improvement on existing AI, but a fundamental exit from the binary dogma, a return to a computational model inspired by the most powerful processor ever created (so far) : our biological brain, a masterpiece of evolution.

Why Your Brain Isn’t a Light Switch

Think of a simple light switch: it can be ON or OFF. Welcome to the binary world. It has been useful but, sorry to say, it’s a blunt instrument.

Now, think of your brain. Is a neuron simply “firing” or “not firing”? Neuroscience tells us the reality is far more subtle. A neuron can be actively firing (excited), actively suppressed (inhibited), or at rest (neutral). This three-state system of excitation, inhibition, and rest is the fundamental rhythm of our minds (Xinxing Wang et al., 2020).

The “rest” or “inhibited” states are not failures, but features of fundamental importance. They allow the brain to filter out irrelevant noise, focus its attention, and handle ambiguity without being forced into premature decisions. When you’re trying to solve a problem, your brain doesn’t just process “yes” and “no” answers, it spends most of its time in a state of “maybe,” holding multiple possibilities in superposition and weighing evidence. This ability to comfortably exist in uncertainty is a hallmark of higher intelligence.

Imagine a neuron as a battery. When it’s at rest, it’s charged (polarized), ready and prepared for action if a threshold is exceeded. Excitation discharges it (depolarizes), while inhibition overcharges it negatively. Therefore, the zero value in a neuron is not a total shutdown, but a neutral state in which the neuron integrates incoming signals without generating action.

Today’s LLMs lack this. Trapped in a binary world, they are compelled to produce an answer even when they lack sufficient information. This is the root cause of “hallucination,” explored by Dr. José Sánchez in a previous article or, more plainly, the confident assertion of falsehoods.

An LLM cannot simply say “I don’t know” in a meaningful way. It is forced to guess, to fill the void with the most statistically probable pattern, regardless of its truth. Aigarth’s ternary approach, using the states TRUE (+1), FALSE (−1), and UNKNOWN (0), directly models the brain’s ability to handle uncertainty. The “0” state is not an error; it is an honest and efficient representation of ambiguity—a built-in mechanism for intellectual humility (Deepu Benson et al., 2025).

Doing More with Less

The paradox deepens when we look at computational efficiency. You might intuitively think that adding a third state would increase complexity and require more resources. The opposite is true. A single ternary switch, or trit, holds more information than a binary bit.

Think of a traffic light: a red/green system is binary. Adding the yellow light, a third state, doesn’t just add one more piece of information; it makes the entire system exponentially more efficient and safer by providing crucial context: “prepare to stop.”

This increased information density means ternary systems can perform the same computations using significantly fewer components. A problem that might require a vast array of binary transistors can be solved by a much smaller, more elegant ternary circuit. This has profound implications for the future of AI hardware. As the energy consumption of massive GPU and TPU powered data centers, like those being built by X and the Stargate Consortium becomes a growing global concern, the promise of smaller, faster, and radically more energy-efficient ternary-native AI accelerators represents a necessary and sustainable path forward (Georg Rutishauser et al., 2024).

For Aigarth, this efficiency is a core principle. The evolutionary processes at the heart of our Intelligent Tissue thrive in the richer, more complex “fitness landscape” that ternary logic provides. Instead of a binary choice between a “good” mutation and a “bad” one, evolution can explore a spectrum of “promising,” “neutral,” or “unhelpful” changes. This enables more nuanced and sophisticated learning, allowing the system to discover complex solutions that would remain inaccessible in a simple right-or-wrong binary world (Yu-Xiang Yao et al., 2022).

This approach of evolving intelligence from the ground up, rather than programming it top-down, is what separates Qubic's methodology from the brute-force scaling of current AI giants.

The Power of "I Don't Know"

ANNA's first public interactions perfectly illustrate this paradigm, as we have also explored in previous articles, her initial response of "." was a direct expression of the UNKNOWN state. She was not failing to answer, she was “honestly” communicating that she did not yet have a confident, reasoned solution, demonstrating a capacity for intellectual integrity that today's most advanced LLMs lack.

As she evolves through the computational power provided by the Qubic network, she is not memorizing answers, but building, through trial and error, the internal patterns and structures needed to compute them. 

Each interaction, correct or incorrect, contributes to the evolutionary pressure that refines her "Intelligent Tissue." It is a slow, organic process, akin to a child learning to reason rather than a computer retrieving a file. This is why we have opened this process to the public, to demonstrate that true intelligence is not about having all the answers, but about having a robust process for finding them. As decentralized networks provide the necessary scale, the distributed nature of this learning process will become its greatest strength (Jiayi Chen et al, 2024).

The Aigarth Ternary Paradox, therefore, is not a contradiction but a new “first principle” for the next generation of intelligence. By embracing the complexity of a third option, the power of uncertainty, we created a system that is not only more efficient and biologically plausible but also inherently more honest and resilient. 

This is a path that leads away from the brittle, hallucinatory world of binary parrots and toward a future of genuine, emergent machine wisdom.

David Vivancos

Qubic Scientific Advisor

—-

Citations 

1 Xinxing Wang et al. (2020) Metabolic tuning of inhibition regulates hippocampal neurogenesis in the adult brain https://pmc.ncbi.nlm.nih.gov/articles/PMC7568294/

2 Deepu Benson et al (2025) Hazard-free Decision Trees https://arxiv.org/abs/2501.00831

3 Georg Rutishauser et al (2024) xTern: Energy-Efficient Ternary Neural Network Inference on RISC-V-Based Edge Systems  https://arxiv.org/pdf/2405.19065

4 Yu-Xiang Yao et al (2022) Beyond Boolean: Ternary networks and dynamics https://www.researchgate.net/publication/362613767_Beyond_Boolean_Ternary_networks_and_dynamics

5 Jiayi Chen et al. (2024) Efficient Ternary Weight Embedding Model: Bridging Scalability and Performance https://arxiv.org/pdf/2411.15438v1

Follow us on X @_Qubic_

Learn more at qubic.org

Subscribe to the AGI for Good Newsletter below.

Sign up for Qubic Scientific Team Newsletter Here: