Hacker News Reader

Comment by draugadrotten | Hacker News Reader

> It is obvious what people want. A single app that does it all. Seriously.

This is the vision of SIRI. Now you know the goal. Please deliver.

Siri's mistake (which also applies to Cortana and Google Now) is to bet on natural language. I seriously believe natural language to be a mediocre interface, and we have the opportunity to do much better. We need to create a completely new language.

This is actually a really interesting point. So we have English (optimized for person->person), standard programming languages (optimized for text->computer), and now you're saying we should make Language X, which is optimized for speech->computer.

I disagree with it, but it's still interesting :)

We need a language optimized for human->computer->human. We already have a few, they're called GUIs. But there's little to no standard surrounding them. Sure, some patterns like "x" to dismiss something or pinch to zoom are widespread, but they don't constitute a complete language.

The binary equivalent is Akinator [1]. It's big on Tinder too.

Basically, you want a language that lets you point at things you don't like. This gets converted to a need, which is then communicated to an array of agents whose job is to come up with solutions. Then enters the flow where you pick a solution (most of which come with conditions, such as completion time or cost), which then becomes a contract that feeds the timeline/calendar/task management system that drives your day to day actions. There's not a lot more to it.

[1] http://en.akinator.com/

Akinator UX is like a marketing survey, a bit tiresome after the first dozen questions with no end in sight.

Yes - natural language is dependent on facial expressions, tone of voice, shared mutual history, shared culture, and so many other things that you don't have, or even want, when interacting with a computer. People can learn a set of imperative commands for interfacing with an AI. People aren't dumb. But the execution has to be good enough to make it worthwhile.

Problem being that you're now relying on the willingness of people to learn your new language, which creates a friction point, like someone here was talking about.

Reply on news.ycombinator.com