That feels like the kind of thing that sounds good on the surface, but breaks down when you actually try to size it. The incremental value of that data vs search data seems nearly 0 - I would guess almost everyone who has searched for a fridge has said some variation of "fridge" out loud, and there would be so much additional noise from people who say it but aren't in any way intending to buy a new fridge.
But don't get me wrong, it is totally plausible that some eng team somewhere would still build it because it's a shiny new tech, it just doesn't seem like the kind of thing that has any real world value.
I feel like you’re overestimating the difficulty of having a word cloud AI trained for relevant phrases. Not just fridge, but (to continue my poor example) “fridge doesn’t keep vegetables crisp long enough”, “wish my fridge had a screen”, “going to renovate the kitchen this summer”, “the neighbours have a better fridge than us, it’s red and shiny”. These are all juicy indicators for possible high-ticket purchases in the current sales cycle.
But don't get me wrong, it is totally plausible that some eng team somewhere would still build it because it's a shiny new tech, it just doesn't seem like the kind of thing that has any real world value.