Other than with language models, this has already happened: Take a look at apps such as Merlin Bird ID (identifies birds fairly well by sound and somewhat okay visually), WhoBird (identifies birds by sound, ) Seek (visually identifies plants, fungi, insects, and animals). All of them work offline. IMO these are much better uses of ML than spammer-friendly text generation.

Here are a few other interesting green automotive startups that didn’t make it:

• Sono Motors’ Sion: Compact EV with solar panels, power sharing, intended to be easily repairable and included a detail manual. They had prototypes but never went to production. Now the company does niche solar applications.
• Workhorse: Series Hybrid (think Chevy Volt) Pickup truck with onboard power for tools etc (was announced around or even before Rivian). Was a very pragmatic idea IMO. Later sort-of resold to Lordstown. Now company does some other things, like drones.
• Lordstown Motors’ Endurance: EV Pickup Truck with hub motors. Made a few hundred, but they have been dragging it out long enough for Ford to make electric pickups. And the idea wasn’t too original even when it was announced.