The LLM is the Interface of the Future
Make computers work for us, rather than us adapting to them.
Imagine telling someone in the 1980s that, one day, they could control a computer with their voice. They’d likely laugh, pointing to their clunky command-line interface. Yet today, over 70% of smartphone users already rely on voice assistants.
Over the decades, how we interact with computers has drastically changed. In my view, the main trend in Human-Computer Interactions (HCI) for the last 50 years has been to remove the distance between user intent and computer action, making our interactions with technology more intuitive and seamless,
In the 1970s, we had command lines. Then, the Macintosh introduced the graphical interface, letting us control computers with a mouse. The iPhone revolutionized this again, replacing the mouse with touch and gestures.
Large Language Models (LLMs) will further close the gap between human intent and computer action. With LLMs, we state our intents, and the software does the rest. This brings us closer to seamless brain-computer interaction, even before Neural Computer Interfaces become mainstream.
Make computers work for us, rather than us adapting to them.
In the future, LLMs will be our primary interface. We’ll manipulate content by talking to an app instead of clicking buttons. We’ll trigger complex workflows with a single command, and we’ll shop by telling Amazon what we want to buy, asking it to compare products, instead of navigating multiple pages and expanding descriptions and sorting comments.
This evolution goes beyond Co-Pilots. Co-Pilots abstract away multiple steps through LLM interfaces, like asking for data from Excel to be inserted into a pitch deck instead of manually copying it. While we’ll see more co-pilots, this is just the beginning. In five years, software will operate very differently.
And it’s not about putting everything on autopilot. AI agents will perform many repetitive tasks, but they won’t replace humans using software. We’ll still need interfaces to manage agents, prioritize tasks, and engage in creative work.
But the way we use software will look dramatically different.
If the computer was once the bicycle for the mind, AI-powered software will be the speedboat.