Today I needed to refresh my memory on what the Boltzmann constant is in units
of wavenumbers per kelvin for reviewing a paper. I used DuckDuckGo to search
(it's 0.69 in case you wanted to know) and the automatic AI assistant on
DuckDuckGo helpfully provided me with the following information:

> The Boltzmann constant can be expressed in wavenumbers (cm⁻¹) by converting its
> value from joules per kelvin. The value of the Boltzmann constant is
> approximately 1.4388 × 10⁻⁴ cm⁻¹·K⁻¹.
That's not what I remembered, so I clicked for more information and got the following.

> Boltzmann Constant in Wavenumbers
> The Boltzmann constant is a fundamental physical constant that relates
> temperature to energy. It is commonly expressed in various units, including
> joules per kelvin (J/K) and electronvolts per kelvin (eV/K).
>
> Value of the Boltzmann Constant
> The value of the Boltzmann constant is:
> In Joules per Kelvin: kB​=1.380649×10−23J/K
> In Electronvolts per Kelvin: kB​=8.617333262×10−5eV/K
> Conversion to Wavenumbers
> To express the Boltzmann constant in wavenumbers (cm⁻¹), the following
> conversion can be used:
> 1 eV is approximately equal to 8065.5 cm⁻¹.
> Using this conversion factor, the Boltzmann constant in wavenumbers can be
> calculated as follows:
> In Wavenumbers:
> kB​≈8.617333262×10−5eV/K×8065.5cm−1/eV≈6.947×10−4cm−1/K

It's OK up to the last step, but the multiplication is wrong by a factor of 1000
and the number is different to what it previously provided. The capitalisation
of units, names and symbols was flawless, however.

I guess it's fortunate there's very little numerical calculation in programming,
otherwise this might be a problem when using them for coding.

@engtao Probabilistic model shit at doing deterministic.

Had one supposedly doing CAD from first principles. I had to explain Pythagoras, as it was trying to do crazy things with something that needed a solution - using Pythagoras.

An then it got in another regression loop, putting all the old bugs back in, so that's when I sat down, and did it all on paper.

@grumpysmiffy @engtao this is really bad. I asked some students in the lab, if they can express Tesla unit in basic SI units. They asked a chatbot, which produced half page of something which “looked” alright, but was wrong (missed coulomb, or put it in wrong place or something) Very insidious