It's not effective on the large fraction of people who tune out due to the length.

The skill is in the fine tuning, probably.

And the not screwing with it while it’s still working hah.

Another hard-won lesson, no doubt.

Like a double negative.

It can get hard to follow after a while :)

How about a triple-play, damning with damn damnation?

It's a thing inherent to diffusion models I think, a kind of uniformity of content across the image.

The similarity may be partly the image model's fault – especially if it's been post-trained/distilled towards performance, correctness and "quality" (for some value of "quality" anyway) which inevitably occurs at the expense of creativity and variation.

But I think it might be more about the LLM's lack of creativity in coming up with the prompt for the image (or "embellishing" a user's prompt), or directly the embedding vector if the LLM is itself the image model's text encoder. Then whatever the image gen outputs is simply an instance of the GIGO principle. It would be interesting to test whether similar cliches and motifs also occur if you ask the model to create an SVG rather than a raster image.

Kind of like analogue TV. Sure it worked and was simple, but consumed some prime spectral real estate.

The other sort of network, ironically.

That's the essence of it, I suppose. Cheap walls of text.

Edit: although another quote comes to mind- "Quantity has a quality of its own."

I remember a company that sold or used bitcoin mining rigs as swimming pool heaters.

There was one that tried to sell them as free heating for the elderly.

It's useful because if an issue is difficult, both levels can disavow responsibility.

in german there is a word for this

Zuständigkeitspingpong

competence table tennis

Sry to be pedantic (it is german after all) but,

"responsibility table tennis" would be a better fit.

Competence in english translates to the skill side of the word

> in german there is a word for this

I am not surprised.

> competence table tennis

Magnificent!

That helps with dogfooding the VM experience, I suppose.