A few days ago, my brother sent me a link to a recent video by Sabine Hossenfelder discussing an MIT paper that claims to build a new bridge between classical and quantum physics. Given some of my own amateur reflections on quantum ontology and particle models over the years, the topic naturally caught my attention and so I felt compelled to take a closer look:

• The MIT press release was, unsurprisingly, ambitious: quantum weirdness may not require quantum mechanics after all. Classical physics, suitably reformulated, might already contain the essence of quantum behavior.
• Hossenfelder’s response was sharp—and skeptical. In the video, she argues that the paper likely overstates its claims and may even contain a circular mathematical argument. More amusingly still, she notes that ChatGPT, Claude, and Grok all apparently agreed with her assessment almost instantly.

That, in itself, struck me as fascinating. So I did what one now apparently does in 2026: I asked “my” ChatGPT (by which I simply mean the instance shaped by years of my own ongoing projects, discussions and questions) what it thought about ‘her’ ChatGPT agreeing with her criticism of MIT physicists. The result was unexpectedly nuanced.

• The AI largely agreed with Hossenfelder that the MIT press release probably exaggerates the implications of the work. Reformulating quantum mechanics using Hamilton–Jacobi theory, least-action principles, path integrals, or hydrodynamic analogies is not entirely new. Such bridges between classical and quantum formalisms have existed in various forms for decades.
• At the same time, the AI also suggested that dismissing the work too quickly may itself miss the point. Reformulations can still be useful even when they do not overturn existing theory. Physics progresses not only through new equations, but also through new representations, computational shortcuts, and conceptual bridges.

But perhaps the most interesting part of the exchange concerned the role of AI itself:

• Large language models are excellent at recognizing patterns, hidden assumptions, familiar forms of circular reasoning, and inconsistencies in argumentation.
• But they are not theorem provers. Nor are they independent judges of truth.
• They are strongly influenced by framing and context. In other words: if one asks skeptically, they often respond skeptically.

That realization feels oddly important. We are entering a moment in which AI systems are increasingly being invoked rhetorically in scientific discussions:

• “ChatGPT agrees with me.”
• “Claude confirms the derivation is wrong.”
• “Grok spotted the flaw instantly.”

Perhaps useful. Certainly interesting. But not equivalent to mathematical proof.

For me personally, the discussion also clarified something else: I do not see this MIT work as confirmation of the sort of speculative ‘RealQM’ or particle-ontology ideas I have occasionally explored over the years on this blog and in open research fora such as ResearchGate or viXra.org.

The MIT approach remains fundamentally mathematical and formal: a reformulation of existing quantum mechanics. The questions that continue to interest me are rather different:

• What is a particle, physically?
• Does phase correspond to something physically real?
• Is there a deeper internal structure or dynamics beneath the formalism?
• Are some of the abstractions of modern quantum field theory descriptions of reality—or merely successful calculational tools?

Those are ontological questions more than computational ones. In that sense, this recent discussion also reminded me of a thought I had while reading Sabine Hossenfelder’s Lost in Math earlier this year.

• Her critique of modern theoretical physics is often presented as deeply anti-mainstream—and in sociological terms, perhaps it is. She sharply criticizes the overreliance on beauty, elegance, symmetry, and speculative mathematical aesthetics. I largely agree with that critique.
• But I increasingly suspect that her criticism still operates largely within the conceptual boundaries of the Standard Model and contemporary quantum field theory. The mathematical formalism itself is rarely questioned at the level of physical interpretation.

My own dissatisfaction lies elsewhere. Not with mathematics as such, but with the possibility that modern physics may sometimes confuse predictive success with genuine understanding. Or, as I wrote in an earlier post inspired by Lost in Math:

“The real challenge is not to extend the mathematical formalism, but to understand what the existing formalism is telling us about physical reality.”

Looking back, this also feels like an appropriate reflection for what happens to be the 400th post on this blog since I started writing Reading Feynman in 2013.

Over time, the project gradually evolved away from the excitement of speculative “breakthroughs” and toward something quieter: trying to reduce the sense of mystery surrounding quantum mechanics without pretending to have “solved” it.

• Not by rejecting mathematics, but by repeatedly asking what the mathematics is actually saying.
• Not by dismissing mainstream physics, but by trying to distinguish between prediction, interpretation, ontology, and scientific storytelling.

And perhaps also by becoming increasingly skeptical of hype in all its forms:

• hype surrounding speculative theories,
• hype surrounding anti-hype,
• and now perhaps even hype surrounding AI-assisted certainty itself.

Modern science communication sometimes oscillates between simplification and debunking, with each side occasionally amplifying the other. Meanwhile, quantum mechanics remains quantum mechanics. And perhaps that is why I found this whole MIT / Hossenfelder / AI-discussing-AI episode so strangely revealing:

• The MIT press office oversimplifies.
• The YouTube critique oversimplifies the oversimplification.
• AI systems then participate in evaluating the critique of the oversimplification.

Interesting times.