Re: I asked ChatGPT to prove that the 1965 muon decay experiment WAS A HOAX.

On Fri, 2 May 2025 3:24:56 +0000, rhertz wrote:

On Fri, 2 May 2025 2:05:39 +0000, ProkaryoticCaspaseHomolog wrote:
>

On Fri, 2 May 2025 1:06:13 +0000, rhertz wrote:
>

But dig into the technical reality, and a very different picture
emerges. The “signal” they claimed to extract was orders of magnitude
smaller than the spectral noise of the gamma ray source—and not encoded
in any way that could be isolated.
Let’s examine why this experiment, widely celebrated in textbooks, may
be better described as a statistical illusion cloaked in prestige.

>
When I was an undergraduate, I worked for a summer in a Mössbauer
spectroscopy laboratory where we *routinely* measured chemical shifts
on the order of 5% of the natural linewidth. Personally, I find it no
stretch at all of the imagination to consider measurement precisions
an order of magnitude or so better than what we routinely achieved.
>
I bounced around a bit in the labs that I worked in. The following
summer, I worked in a lab doing split brain research, and the summer
after that, in a lab studying the effects of GABA on crayfish neuron
signal transmission.

>
>
Quite a difference between 5% shift in natural bandwidth (you) and A
NOISY 0.1% shift (Pound allegation, 1959-1960). I think that ChatGPT
explained this disgraceful event with astonishing clarity:

No, it really ISN'T that much when you consider the differences in what
we were attempting to achieve. We were interested in measuring the
OVERALL chemical shifts including hyperfine splitting and quadrupole
splitting in the analysis of Moon rocks. The spectra were complex since
the rock samples were complex mixtures.
We were decades away from having technology to analyze individual grains
of the rock samples. Many of the conclusions that we reached in the
1970s have needed to be heavily revised in the light of later, vastly
more refined results.

"this experiment, widely celebrated in textbooks, may be better
described as a statistical illusion cloaked in prestige".

As I've stated above, it is not at all hard to "jailbreak" an AI
using adversarial prompts to get it to spit out nonsense reflecting
nothing more than your own prejudices and misconceptions.

I always called it a farce, an HOAX. And that's what it was, is and will
be.