Inspired by 'Her': 7 AI interactions we'd like to see in the world

We’re still in the awkward, early phases of AI, where the technology is so new, the only way people have figured out how to interact with it is a way they’re already familiar with—even if it doesn’t make the most of the new technology. (ChatGPT itself admits its interface wasn’t designed with users in mind.)

Given where we are in the evolution of the technology, we decided to imagine new models for how we would want to interact with A.I. We borrowed a little inspiration from the Joaquin Phoenix movie Her, about a man who falls in love with the operating system in his phone, and the concept of calm technology, which is unobtrusive and embedded in our everyday lives, to come up with seven ideas for AI interaction paradigms we’d like to see in the world. There was only one rule: no chatbots. (Spoiler: We broke the rule.)

Here’s what we came up with.

1. Neverending Story Book: AI-generated bedtime stories 

This hardware device uses generative AI to always tell a new tale—and one that users can help shape. While some inputs are constrained, like the genre knobs, the subject matter open input box can accommodate as much creativity as the user can produce. This combination of constrained and flexible inputs provides guardrails for how users are meant to interact with the device, and by extension generative AI.

2. Tamaglitchi: An AI pet, embodied

It’s about time somebody made a pet with an off button. Tamaglitchi uses models trained on thousands of hours of dog and cat noises, behaviors, and interactions with humans to create an experience that is so close to having an actual pet, your parents can’t say no. (A prototype to meow at, via Character.AI.)

3. Portals: Handheld, AI-enabled circular lens that enhances connections

With so many devices that command our attention and pull us away from our physical environment, Portals are windows that allow us to refocus on the people in our lives and the natural world. Leaning into the passthrough quality of augmented reality, Portals allow users to capture memories and style inspiration in real time with minimal disruption to their field-of-view. Using machine learning models, Portals provide users with valuable social nudges and additional context to objects, people, and places in their environment.

{{video-1}}

4. RemixRemote: Custom entertainment content on demand

The future of personalized entertainment: When you have a specific amount of time available and an inkling of what you might like to watch, you can generate something that meets your specific needs. The RemixRemote uses generative AI to create unique content tailored to your taste and time availability. Collect ideas of what you’re interested in, then press down on the voice input dial to describe qualitative aspects of your inspiration. Pressed for time? Each turn of the dial cuts (or adds) 15 minutes to your content’s duration, making sure you can fit this into your schedule.

{{video-2}}

5. Collective Sentiment Mirror: An installation that summarizes conversations 

This AI will gauge the conversations in a room and spatially map them in an installation in the ceiling. Each person is represented by a glowing dot, and as they interact with other people, interconnections between the dots grow. Opt-in poetry: The ceiling can also generate poetic summaries, picking up nouns or a single word based on the conversations happening in the space. 

{{video-3}}

6. Scent Capture: Meet the eNose

eNoses digitize odors using a graph neural network. Once scent is recorded, scent capture will organize, tag, and combine smells to create AI-generated scents as unique as you. Relive your experiences by sending your custom scent profiles to be translated into physical forms like perfumes, candles, and essential oils.

And for the last one, we tossed out the rulebook, leaning into chatbots.

{{video-4}}

7. ArticulateGPT: An LLM that reduces articulation barriers

ArticulateGPT seeks to address one of the biggest issues users face today: their ability to translate their thoughts into words. It gives users options for what to say, so they don’t have to deal with the paralysis of a blank slate. But it’s not just to help with speech or written communication; users can flow between multiple media types, like text, code, imagery, video, audio, and even 3-D. 

How do you want to interact with A.I.? We’d love to hear from you: ai@ideo.com

Special thanks to Grishma Rao, Neal Boyer, and Ceci Gomez for their work in coming up with these concepts.