Asking an assistant to do an action is like hitting that "I'm feeling lucky" button.

For example, imagine asking Alexa to buy you something.

You're basically saying, "do a search for that topic, click the top result, and buy one."

This implies that you really trust the quality of that search result to give good results and have the best option as the first position.

How willing you are to take this leap of fate is tied to:

1) How good do you expect the search results for this query to be

Based on how good the search results in general have been in the past

As well as how good you expect the service to do on this query in particular.

2) How big the downside for a miss is

If you're buying a $100 item that will be shipped to you, the downside is quite large; you're out $100 and only realize it was wrong days later.

If it's a Google Search, the downside is tiny; just immediately do the query again manually.

Amazon's search results have famously become quite scammy and crappy, which means trusting Alexa to buy an item for you has gotten harder.

The common AI assistant use case of "plan a trip for me and automatically buy tickets" is almost impossible to imagine working in practice.

Both because it's hard to imagine high enough quality

The preferences we have for travel are subtle and hard to explain to a human we know well, let alone a machine.

And because the downside cost is so high.