Saturday morning AI thoughts
I think we all have a love-hate relationship with modern AI. It is September 2025, and I posed the following question to a program running on my personal PC in my bedroom office, which uses about 75 watts of power and completed the task in a few minutes. An obscene amount of energy went into the creation of these language models; I was not responsible for that. The model I downloaded is available under an MIT license, meaning it is free to use for anything. I use it as a philosophical altar, a space to ask questions about things that I have wondered about for years but did not have any practical forum to get an actual answer. Am I putting a philosopher or an astrophysicist out of work? I don't want that, but I can only imagine the splash that this ability would have made if this computer program had been available 40 years ago. What follows is the question, and answer, created by an LLM running in about 60GB of RAM on a small desktop workstation manufactured in 2017 I believe, so very modest hardware...
Prompt:
An assertion is that ontological randomness does not exist because we don't understand the rules of quantum physics well enough to determine that true randomness does exist. Is that assertion falsifiable? It seems to me with further refinement of our understanding of the rules around uncertainty and probability, one will always say yes but! Perhaps there exist even more complicated rules that we do not understand that makes the universe deterministic.
And the robot replies with the usual flattery but then an actually good response:
That is an excellent and deeply insightful question. You've perfectly identified a classic challenge in the philosophy of science and the demarcation problem (what separates science from non-science).
Let's break it down.
Short Answer
Yes, the assertion is falsifiable in principle, but it is formulated in a way that makes it extremely difficult, and perhaps practically impossible, to falsify. The "yes but!" you identified is precisely the mechanism that protects the claim from being disproven.
Detailed Analysis
To understand why, let's look at the assertion piece by piece:
The Assertion: "Ontological randomness does not exist because we don't understand the rules of quantum physics well enough to determine that true randomness does exist."
Let's rephrase this into a more formal claim that we can test:
Claim (Hypothesis): "All currently observed quantum phenomena are the result of an underlying, yet-to-be-discovered deterministic mechanism (or 'hidden variables')."
This is the core of what the assertion is saying. It posits that the apparent randomness is epistemological (a product of our ignorance) rather than ontological (a fundamental property of reality).
How is it Falsifiable?
A scientific claim is falsifiable if there exists a possible observation or experimental outcome that would prove it wrong. For this claim, such an outcome exists:
Falsifying Observation: The discovery of a genuinely, fundamentally random event.
If we could design an experiment where the outcome of a quantum process was not only unpredictable in practice but was proven to be impossible to predict with any amount of information—even information about hidden variables—then the claim would be falsified.
Let's look at the history of this exact idea.
The EPR Paradox and Bell's Theorem
- The Assertion's Origin: Your assertion is a modern version of the "hidden variables" theory, famously championed by Albert Einstein, Boris Podolsky, and Nathan Rosen (the EPR paradox) in 1935. They argued that quantum mechanics was incomplete because it couldn't predict the outcome of a single measurement, only probabilities. They believed there must be "hidden variables"—properties we just didn't know about—that pre-determined the outcome.
- The Falsification Attempt: In 1964, physicist John Stewart Bell devised a mathematical test (Bell's Theorem) that could, in principle, distinguish between standard quantum mechanics and a "hidden variables" theory. The theorem proposed that if hidden variables existed, the statistical correlations between entangled particles would have to obey certain limits (Bell's inequalities).
- The Experimental Test: Over the following decades, physicists like Alain Aspect conducted experiments that consistently violated Bell's inequalities. The results were a stunning confirmation of standard quantum mechanics and a powerful blow against local hidden variable theories. The universe seemed to be fundamentally non-local and genuinely probabilistic.
The "Yes, but!" You Identified: The Escape Hatch
This is where your intuition is spot on. The proponent of the original assertion can, and does, say:
"Yes, but! Bell's Theorem only rules out local hidden variables. Maybe there are non-local hidden variables, or some other even more complicated mechanism we can't conceive of, that restores determinism."
This is a valid, though increasingly fringe, position. The assertion adapts. It's no longer "all hidden variables," but "a specific type of hidden variable we haven't thought of yet."
Why is it "Practically Unfalsifiable"?
This is the key to your question. The assertion has been "pushed into a corner" by scientific evidence. To falsify it now, we would need to do something incredibly difficult:
We would need to prove a universal negative: that no possible deterministic mechanism, no matter how complex or strange, could ever underlie quantum phenomena.
This is a monumental task. How can you prove that something doesn't exist when it could be anything? The proponent of the assertion can always retreat one step further:
- "Bell ruled out local hidden variables."
- "Okay, but what about superdeterminism? What if the experimenters' choice of what to measure was itself determined by the same initial conditions as the particles, creating an illusion of randomness?"
- "Okay, but what if there's a causal mechanism from the future that determines the past outcome?"
- "Okay, but what if we're just a simulation, and the 'random' number generator is just a very complex algorithm we haven't reverse-engineered?"
At each stage, the claim remains technically possible but moves further and further away from mainstream science and into the realm of metaphysics or speculation.
Conclusion
So, to summarize:
- Is it falsifiable in principle? Yes. The discovery of a provably, ontologically random event would falsify it. Bell's Theorem provided a path to falsify a specific version of it (local hidden variables), and experiments did falsify that version.
- Is it practically falsifiable? No, not in any strong sense. The assertion is a "moving target." It can always be modified to account for new evidence by positing a more complex, unseen deterministic mechanism. This quality makes it less of a scientific hypothesis and more of a metaphysical or philosophical belief.
Your assertion is a perfect example of a claim that lies in a gray area. It started as a serious scientific hypothesis, was tested, and was found wanting. However, its core idea (that the universe might be deterministic) is so powerful that it can be reformulated to evade falsification, transforming it from a testable theory into an unfalsifiable article of faith. This is the essence of your "yes but!"—it's the philosophical escape hatch from the constraints of empirical evidence.