Again, I don’t see chat or voice interfaces replacing other UI paradigms, but as NLP improves and the chat/messaging/voice platforms improve their capabilities, you’ll access services across a spectrum of different contexts. Does it make sense to use a visual interface when you’re driving? Does it make sense to use a voice interface when you’re taking notes in a lecture?
I find that asking Siri to add things to my to do list is more efficient than pulling out my phone, finding my task app, and adding it by hand. But when I need to see what I need to do, a visual list to is much easier to review.
I find asking Alexa for the weather on my way out the door is more convenient than pulling out my phone as well. But when I’m planning what I need to wear for a multi-day trip in a foreign city, it’s easier to look to the web for an extended forecast.
I can imagine that asking the UPS bot to track a package and notify me of updates via messaging will be better than constantly having to fill out a web form.
So, I agree — utility is sparse today, but that was true of the early web as well. I’m confident the value of bots (in some configuration) will prove out over time.