Determinism vs Probabilism is an extremely nuanced discussion that, despite extensive research, still lacks a clear-cut answer or solution. We’ve labeled various systems as “probabilistic” – including LLMs, Markov Chains, Bayesian Networks, and various concepts across different fields, calling them “Probabilistic Systems.” However, how can a system be truly probabilistic if it follows mathematical and logical rules? How can systems that use deterministic procedures to imitate probability be considered genuinely probabilistic?
I believe probabilism is a fallacy. Currently, only biological systems might possess the true ability to be probabilistic (there might be determinism underneath, but due to our limited knowledge of current biological systems, it’s hard to understand how). Everything else is a deterministic system – computers operate deterministically, and if your code doesn’t work as it should, that means you made an error in your code; computers don’t roll dice.
People label LLMs as probabilistic systems, but I believe they’re the perfect example of probabilistic imitation. People discuss that if you put a prompt into an LLM twice, it gives you different responses, but they aren’t really paying attention. Try typing a prompt to an LLM five times – it might return different words, but the whole logic is still coherent, and it’s the same patterns that appear every time. I believe if you try this 100 times, you would see the same effect.
The illusion of probabilism in these systems stems from their complexity and our inability to track all variables involved. When we examine computer-generated “random” numbers, we find they’re actually pseudo-random, generated through deterministic algorithms. The same applies to machine learning models – their apparent randomness is simply a product of complex deterministic processes we haven’t fully mapped out.
Anything called probabilistic is just a euphemism telling us that we don’t understand the rules of how it works. This applies to everything and possibly biological systems (but I would still call it probabilistic for now). Things like Quantum mechanics, which we claim have probabilism, is just cope that we don’t understand the rules and dynamics of the system. As Einstein said, “God doesn’t play dice with the universe.”
Even in cases where we observe seemingly random behavior, like particle physics or neural firing patterns, we might simply be witnessing the effects of deterministic chaos – systems so sensitive to initial conditions that their behavior appears random to our limited measuring capabilities. The appearance of probability might be nothing more than our way of mathematically describing patterns we can’t yet explain through deterministic means.
What we cannot describe we label it probabilism but most systems we create we can call most of them deterministic, cause we humans are the ones which invented it and we know the rules and systems behind it such as computers, a computer might appear probabilistic to a less advanced specie and that is understanding of why biology appears probabilistic to us, the created and programmer of biology understands the rules and processes that cause the emergent behaviour and occurrences in biological systems but due to our limited knowledge of what actually goes on nature, we still understand the rules cause we are looking at it at an extremely low level, maybe if we try to grasp it on a high level we would be able to make some insights about nature itself.
The distinction between deterministic and probabilistic systems may be more about our understanding than reality itself. What we label as probabilistic today might simply be deterministic systems whose rules we haven’t fully grasped yet. This perspective challenges us to question whether true randomness exists in any artificial system, or if probability is merely a human construct to deal with complexity and uncertainty. Until we can fully understand the underlying mechanisms of biological systems and quantum phenomena, we might continue to use probabilistic models as useful approximations, while recognizing that they might be masking deeper deterministic processes. This understanding has profound implications for how we approach system design, scientific research, and our philosophical understanding of causality and free will, but before we step any further, we need clear definitions of what is probabilism and determinism.
rustian ⚡️
