What you are describing would be great for chain restaurants like Chile's or Olive Garden, but for something sufficiently high end the people involved really matter. If you shift your view up a few notches there are many places where real intelligence and dynamicness is required that we can't yet automated.
Having a wine expert or a unique expert chef as part of the experience is part of high end dining. Consider one of those mid to high end Japanese steak houses where there they throw shrimp across the room into someone's mouth. Or a more midwestern steakhouse where you can talk with a person who butchered the steak. Or one of the new microbreweries where can can talk with the brewer about some new experiment has has and look at the brewing equipment right there. None of these make sense to automate, until we get strong AI.
Figuring out suggestions based on preferences and given choices with so constrained options is an easy problem.
The "expert" is there to sell you a story. Not that there's anything wrong with that - the story is a part of the experience that many people like - but let's not pretend it's a technical problem that's difficult to solve.
I know a bunch of people who work in that buisines (my brother is a chef, gradeschool friend as well, ex. is a restaurant waiter, met a bunch of their co-workers, all worked in Michelin star restaurants etc.) what actually goes on vs what the customer hears ...
> Figuring out suggestions based on preferences and given choices with so constrained options is an easy problem.
Then why do Spotify and Netflix et al persist in identifying artists and songs I've heard of (because of obvious similarities to stuff I like) but have actively ignored (because of any of a number of reasons that don't boil down well into an algorithm), over and over and over again?
I still don't want to watch The Office, Netflix. I won't in another six months, either.
But I could tell a person why not, and they could factor that into their recommendations. It's like the difference between trusting Waze and trusting a local that says "even though the app says go another half mile then turn right, turning right here is way easier, since if you wait you'll have to then quickly make your way across three lanes of traffic in a hundred yards to make your left, and making the right here gives you more time to get over."
Different scale/complexity. You could hand-code waiters decision tree for recommendations and the possible choices end in ~hundreds. Personalized music suggestion has to pick from millions of choices and there's no straightforward algorithm behind it.
Chili's is already starting down this path. At the location near me you can order drinks and pay through a device at the table. They haven't jumped in all the way as food orders are still through a server.
Though, as someone who has had a credit card stolen via gas pump several times, I'm really hesitant to use a card at one of these devices. They're probably not very secure to begin with and then you couple that with the fact that anyone can sit with one of them for an hour or so with no scrutiny, it's just ripe for exploit.
Having a wine expert or a unique expert chef as part of the experience is part of high end dining. Consider one of those mid to high end Japanese steak houses where there they throw shrimp across the room into someone's mouth. Or a more midwestern steakhouse where you can talk with a person who butchered the steak. Or one of the new microbreweries where can can talk with the brewer about some new experiment has has and look at the brewing equipment right there. None of these make sense to automate, until we get strong AI.