Maybe I’m misunderstanding, but isn’t generating UI just-in-time kind of risky because AI can get it wrong? Whereas you can generate/build an MCP App once that is deterministic, always returns a working result, and just as AI native.
With this you build your own React components and register them with the AI. The AI chooses which to use and what props to pass into them, so it's not generating UI from scratch, if that's what you mean - other Michael from the Tambo team
I see. how often does AI get the components wrong? This is my big question wrt just in time generative ui. UI built beforehand (also likely generated by ai but vetted by eng team) feels like it has the win here. Why take on risks of non determinism if you don’t have to. This feels like the big pro for mcp apps.
I found this reddit comment. It may be that West African cocoa is all conventionally grown (by a cartel) while American cocoa, where heavy metals are more common, is more likely to be organic.
You are falling for a scam. The truth is that West Africa retains a monopoly scaring you consumers with talk of “lead” and “cadmium”. To begin with, Cadmium is not heavily present in West Africa. It is very common in volcanoes soil like Guatemala, Ecuador, Peru, Venezuela, and Colombia. That said, this fear is conveniently promoted by big chocolate companies because their cocoa is all from West Africa. As for “lead”, this is from pesticides or private aviation. That said, cocoa pods are thick and it is questionable how much of this actually gets into your final product given the thick skin of the cocoa pod. Our concerns should be more focused on the poor farmers many of which are children that get sprayed or exposed to the pesticides and not so much the end consumer in this case. You might want to buy better chocolate. That Lindt is definitely west African cocoa and it is conventional so it gets sprayed. You don’t need to buy organic to get pesticide free chocolate. You just need to buy chocolate from chocolate makers who work with small family farms.
That seems not entirely persuasive. In particular:
> As for “lead”, this is from pesticides or private aviation. That said, cocoa pods are thick and it is questionable how much of this actually gets into your final product given the thick skin of the cocoa pod.
The commenter might have been discussing a different study, but this one appears to have tested the final product. If it found lead, people are eating lead.
The rest of the comment is just not what I care about. They're telling me the companies raising the alarm about heavy metals are doing so out of commercial interest. Okay, but that's just an ad hominem argument. It doesn't mean they're wrong.
I'd rather hear "actually, these levels of cadmium, arsenic, and lead are well within safe limits unless you're eating nothing but chocolate all day" (with matching data of course), which would better reassure me about the fancy South American organic chocolates. And/or if the commenter is trying to say that the more mass-produced West African conventional chocolates are worse, what are the specific health dangers (to the workers, apparently) posed by their pesticides that I should actually be concerned about?
In eggs this is also a problem. The chickens are required to eat natural foods but in certain cases the natural foods contain way more bad chemicals than if they were simply synthesized and added as a food additive to their regular food. Which is how it's done for regular eggs (because it's cheaper).
In the sense of just completely making up references out of nowhere? Very low. Humans make mistakes, of course, but they tend not to be so egregious. Grounding a submission on a made up reference is quite likely to be fatal to your case in a way that most human errors aren't.
This is the important bit that's always missing in discussions about LLM/AI applications to existing industries - what is the rate of mistakes for a human worker?
Maybe I’m misunderstanding, but isn’t generating UI just-in-time kind of risky because AI can get it wrong? Whereas you can generate/build an MCP App once that is deterministic, always returns a working result, and just as AI native.