Leading AI Expert and CEO Explains Why A Change In Our Approach to AI Can Maximize Business Results How we engage with computer colleagues needs to be reimagined in the age of AI. Here's how.

Thoughtless user interfaces can be maddening. A simple task — editing photos, say — goes sour when I'm overwhelmed by navigation bars and pointless pop-ups. Maybe I turn to a Help window, and (sigh) now I'm telling my color correction problem to a glorified FAQ. Only one thing's clear — the software has no idea what I want.

But AI changes that — or, at least, it should. As the CEO and co-founder of an AI-powered customer service platform, I've given this problem a lot of thought. And I believe we're still designing software for the past instead of reimagining it for the future. We're still thinking of our devices as brainless machines — a car you have to sit down in, buckle up and drive. Nearly overnight, that dynamic has changed.

We're not driving a car any longer — we're saddling up a living, breathing horse with a mind of its own — and it's ready to ride.

Related: I Teach AI and Entrepreneurship. Here's How Entrepreneurs Can Use AI to Better Understand Their Target Customers.

Evolving interfaces

Every leap forward in computing creates a paradigm shift in the user's experience. The dawn of personal computers led to the graphical user interface (all those folders and icons you click on with your mouse). Mobile phones led to multi-touch techniques and app-centered home screens.

AI will shake things up again. But how? Will it lead to voice-based interfaces, where we simply talk to our computers? Spatial awareness, with software integrated into your environment through augmented reality? Will it stay mired in prompt writing, where we simply fill our fields to generate what we want? The debate is just getting started.

But here's what I do know: Going forward, the best software will accommodate two intelligences (yours and the AI's). Its core function will be to facilitate a genuine back-and-forth — a collaboration between two minds, each with its virtues and limitations. And it won't look much like the software applications we know and use today.

Most software up until now responds to your commands but doesn't have any ideas of its own. That design paradigm assumes that you're the expert, and the software is just a way to convey your expertise. It makes the annoying assumption that the user knows what they're doing. I think that's dated, and it's got to change.

AI-native software is fundamentally different and in ways that are stranger and more powerful than we've appreciated. I'm not talking about "co-pilots" here — the ubiquitous assistants that have popped up, powered by AI, to whisper advice to you like some grandchild of Clippy. I'm talking about genuine co-workers that anticipate your goals and guide you forward in bespoke ways. Ethan Mollick, a Professor at Wharton, has called this "co-intelligence."

The future Mollick imagines could produce a user experience that's radically different for each of us. Just as humans have foundational personalities and then emphasize different aspects depending on who they're with, AI-native software will have a foundational design but customize itself to meet the user's needs. In its most advanced form, it will literally create a new user interface on the fly based on your history and goals.

Think of that personalized interface as your own custom-made saddle. As any equestrian will tell you, horseback riding is a relationship between two minds, and that should be the goal with an AI interface, too. Your input matters, of course. But so does the energy and intellect of the platform. An ideal UI will allow for the subtlest communication between horse (the AI) and rider (you). It will enable the AI to respond to your style, intuit what you want, and ultimately take you where you need to go.

Related: You Can Fear It and Still Use It — Why Are So Many American Workers Shy About AI?

A sneak peek into two-intelligence design

Here's how our experience with computers should feel when we're riding in an AI saddle:

For starters, forms of input will expand beyond clicking and typing. AIs will see what's onscreen, hear what you're saying, and track where your eyes are focused to better understand what you're trying to do. In my industry — customer service — this change will massively streamline the experience. A large amount of time and energy is wasted when customers have to explain what's happening on their screen (log-in errors, order issues, etc.). Imagine all those hassles being instantly handled by an AI agent.

At the same time, AI will tap user information to understand your skill level, preferences, and goals. They'll use that data to rebuild their own interface for each user. Customized landing pages and recommendation algorithms were just the beginning. Going forward, facial recognition technology could monitor your expressions and trigger extra help when you're frustrated or ease off when work is flowing.

Meanwhile, the notion of "user" itself will blur, and more space will be made for non-human entities. People using ChatGPT are already used to having an Avengers-style team of helpers at their beck and call (called GPTs), each with an avatar like any other remote worker. Going forward, Salesforce, WhatsApp, and all kinds of platforms will host AI agents right alongside humans — to the point where we'll treat them both as friends and coworkers.

To lower our cognitive load, it'll make sense for some of these AI assistants and agents to be explicitly personified: Devin, marketed as the world's first AI software engineer, has his own online workspace and chat window, for example. And then, in other cases where human input is less important, they may work quietly behind the curtain.

That invisible labor will also mean far fewer tools and toolbars in the software that humans interact with. It's the end product that we're after, not the opportunity to toil. User interfaces will grow simpler and more streamlined — places to peruse and learn from finished work, rather than do work, itself. Indeed, venture capitalists are observing that AI start-ups are pivoting to sell finished work rather than software.

Likewise, as AI draws what it needs from apps and collates material for users, there will be even fewer reasons to tap on the most fundamental tools of our digital lives — the apps on our phones and tablets. Someday soon, the app-based home screens on your phone may need to be redrawn.

Nobody can perfectly imagine the future. But I do know that getting the AI revolution right means designing user interfaces with all these components in mind. It's a "horse riding" moment. Your intentions and the AI's work in tandem. And it's made possible by a user interface that makes room for all the benefits that a second intelligence can bring. Giddyup!