Apr 1, 2024

Apr 1, 2024

The future of Conversational UI

The future of Conversational UI

Everything is an input and everything is an output.


AI’s like my 4-year-old nephew. Every week, he wants to be something new when he grows up…

One day it’s a soccer pro. The next day it’s an astronaut. Now, he just wants to be a garbage man.

AI’s similar. It has a ton of different narratives right now.

Human clone. Stalker. World domination. You name it.

Here’s exactly where we are today:

Conversational UX / Chat-styled interactions is what everyone’s making.

Some tasks which are possible through conversational UX —

  • Fire-and-forget-tasks like “play music”.

  • Specific trivia queries like “weather”, adding To-Dos.

  • A conversational partner like an AI girlfriend.

But there are many problems with conversational UX —

  • People land on an empty screen. They then try to decipher what all can be done.

  • People use apps which keep track of their state.

  • Editing — Whether it be video, audio or article, you need store the draft version to come back to later on.

  • Travel planning — tracking what places you’ve seen and which bookings you’ve already made.

  • Researching — opening 50 tabs to keep track of the different directions you’re exploring.

So the next question is obvious — what’s after ChatGPT? Are we meant to be prompt-designers/prompt-engineers/prompters??

Here’s where we are headed, in 2030 :

There are 4 AI innovation trends that are taking place under our noses —

  • Dynamic Interfaces,

  • Ephemeral Interfaces,

  • aiOS ← today’s post,

  • Screenless UX

The other three trends for another time 😉

A look into aiOS — What is it?

aiOS whispered into your ears giving you goosebumps ;) — image source: author

There are many definitions of the term ‘aiOS’, but the most basic one is — an Operating System powered by AI.

Seems obvious right?

Jordan Singer, doing somethin’ somethin’ with AI @Figma, described it as a UX controlled only by conversations.

But conversations is just one medium.

There can be other ways of interacting within the aiOS.

The pull-to-refresh type intuitive interaction is still TBD.

Irrespective of the interaction, the underlying values for the aiOS are going remain the same. Let’s dive into the 4 major aiOS values

1. You don’t go out, it comes to you

Us to the internet before AI — “I will find you, and I will .. “

It’s about bringing everything to you, as a user.

  • At app level — it can be through chatbots,

  • At inter-app level — it can be through Adept, ideally just explaining in a chat what you want to be done, the AI does it for you.

  • At browser level — it can be through Arc Search, you just search and the browser browses for you.

  • and now zoom out further, how would it look at OS level?

  • and now zoom out further, how would it look at the hardware level? (Screenless UX)

2. Interoperable apps

Apps which can communicate between each other

Let’s say you’re a freelance copywriter starting your ⛅️ Monday morning.

> You start by listening to a podcast that you’d scheduled last night.

> You take notes on the side.

> You open your emails ready to send out an important email to you client.

> You leave the email mid-way to get a coffee to freshen up.

> You open your calendar to put in some time with another client.

> You pause the podcast.

> You open your email app to continue writing the mail.

> You get a notification on Teams. You respond with a file.

> You respond again with a link to the file.

It’s lunch time.

Phew, a lot of switching between apps. Now, what if you could browse ALL your things in ONE fluid interface?

The answer? Itemized workspaces.

All apps are items or features.

You can drag drop your podcast episode into your notes app. Not as a reference, but the episode itself. You can drag-drop your half written email into your notes to come back to again. You can drag drop the flight you want into your calendar and it’s booked.

Any app/item can be pulled into any other app/item.

It’s all intuitive, much faster and clearer.

2.5. Built-in OS level solutions

Bringing apps to OS level has been happening since the beginning of time. When the app store was launched, basic features were stand-alone apps.

For example, the flashlight apps.

Flashlight apps in the App store vs Now in the OS

Similarly, Grammarly or ChatGPT that helps us write better (with auto correct or text prediction) need not be at app level right? It could easily be at OS level, built into the keyboard.

3. Context is foundational

The problem with current AI applications (read: conversational AI-UX) is that they aren’t in the same context as the user.

In MS Excel, the chatbot doesn’t have the complete context of what you’re working on, what’s your working style or even, your completion deadlines.

A screenshot of MS Excel’s AI chatbot

A simple application of this in the traditional setting (apps and websites) would be — what if websites had the context of how many times you’re visited?

You can adapt the UI based on the visits.

a concept by @AntoleOis on X — image source: @AntoleOis

Now imagine scaling this at an OS level.

A good example of this is what if your input method is determined by how you’re positioned with the device?

  • > If you’re looking at your laptop, the input is via keyboard.

  • > If you’re looking away or standing away from your laptop, the input is via audio.

A lovely demo by the cofounders of New computer shows just this!

2 states of the screen depending on where user is looking — Image source: AIE Summit

Adding context of the user from outside the bounds of the app/website makes the user experience much more intuitive and faster.

4. Everything is an input

You might have guessed this one. It overlaps with the above value.

AI has hacked the Operating system of a human being — language.

~ Yuval Noah Harari

And because AI can understand language, essentially conversations, it also understands all the mediums of communication i.e., voice, visual and text.

So now everything is an input and everything is an output.

You can input in text and get an output as a visual — without you have to choose if that’s the best medium, ai does that for you.

Have you checkout the ChatGPT’s voice read-aloud feature? It. is. so. freaking. real 🤯 It pauses, breathes, and speaks just like a human. You gotta try!

And that’s it, those are the 4 values being considered to create the aiOS. So what do you think..

…is AI powered OS the next big thing?

On a completely random note, if aiOS was a movie

" Solid overview of emerging design
patterns and well organised content

" Solid overview of emerging design
patterns and well organised content

Most evaluations of the trends have been pitiful.
[But] the AIverse Design team nailed it! "

Most evaluations of the trends have been pitiful. [But] the AIverse Design team nailed it!

Jacob Sullivan
CPO at Faculty.ai

Jacob Sullivan
CPO at Faculty.ai

Get the 'Trending AI-UX Patterns' ebook

© 2024 Aiverse. All right reserved.

Designing for AI, Augmenting with AI

© 2024 Aiverse. All right reserved.

Designing for AI, Augmenting with AI