I remember reading a report where people running AI-model instagram account were using insanely long and detailed prompts about the setting, lighting, makeup, pose, disposition, clothing, etc. about their models. Presumably with some reference image of the face / body to remain consistent across images.

It‘s not clear to me whether a sufficiently detailed prompt can generate actually interesting video with a natural ”texture” (for lack of a better word).

I remember a paper which was posted on HN a few weeks ago where somebody implemented KAN networks in FPGAs, since those can readily be approximated as LUTs.

And even at that point, others will continue to write using AI to simulate being one of those people.

Like a big-boy prius.

Here's the key point in the article: A month ago, the bond was priced at +175bps yield above treasuries with similar maturity, now it's priced at +231bps.

I believe his family owns 2 cars, but had to replace one of them recently when a pebble caused a freak accident, essentially totaling the engine in one of the cars. Hence 3 cars purchased in 10 years.

Edit: Episode #592, at the very end: https://overcast.fm/+ABQnv4mYCFc/1:40:38

50% of all energy in the swiss economy is oil / gas. Of the remaining 50% (electricity), 2/3 are generated by hydro. The remaining ~1/3 by nuclear fission.

Swiss electricity prices are sky-high, and the demand for electricity is going to continue to rise.

To remain a competitive industrial economy, to transition away from oil/gas, and to offset any potential losses of hydro power as glaciers melt, nuclear + solar is the only real path for switzerland.

Physics.

As a system provider you take responsibility not only for maintenance, but also to tank the blame once something invariably goes wrong