It’s not out of the question that we get emergent behaviour where the model can connect non-optimally mapped tokens and still translate them correctly, yeah.

It is a concern.

Check out https://tiktokenizer.vercel.app/?model=deepseek-ai%2FDeepSeek-R1 and try entering some freeform hexadecimal data - you’ll notice that it does not cleanly segment the hexadecimal numbers into individual tokens.

Still, this does not quite address the issue of tokenization making it difficult for most models to accurately distinguish between the hexadecimals here.

Having the model write code to solve an issue and then ask it to execute it is an established technique to circumvent this issue, but all of the model interfaces I know of with this capability are very explicit about when they are making use of this tool.

Is this real? On account of how LLMs tokenize their input, this can actually be a pretty tricky task for them to accomplish. This is also the reason why it’s hard for them to count the amount of 'R’s in the word ‘Strawberry’.

Without actually knowing how much constructing the physical buttons cost, I would guess that the real savings are in process optimization - if all you have for the interface is a screen, then you don’t need to have the interface design done before constructing the car - you can parallelize these tasks.

Insufficient as far as justifications go, but understandably lucrative.

The caveat there is that you’re personally responsible to maintain backups of your image catalog when self-hosting.

Same, but also apps