Skip to main content

How Future Digital Interfaces Will Combine Screens, Gestures, Voice, and AI

Published on November 15, 2026

7 min

A beautiful blonde relaxing in a mountainous setting
  • Interfaces are evolving toward greater multimodality: natural language, gestures, eye tracking, and touch will redefine how we interact with machines.
  • Four scenarios are emerging for 2035: augmented screens, ambient interfaces, AI agents, and nearly invisible interactions.
  • Accessibility, data protection, and user comfort will determine the success of these new interfaces.

From keyboards and mice to touchscreens, interfaces have gradually simplified our access to digital technology. Today, our interactions with machines are diversifying and becoming increasingly intuitive by incorporating gesture and voice recognition, eye tracking, and augmented reality.

Recently, a new milestone was reached with the rise of generative artificial intelligence, as users can now directly express their intent rather than searching for a function within an app. The system interprets the request, mobilizes the necessary resources, and provides a context-appropriate response.

By 2035, the progress made by these technologies could profoundly transform the way we browse the internet. We explore this topic with Emmanuel Pietriga, research director at Inria (National Institute for Research in Digital Science and Technology, France) and a specialist in human-machine interaction, data visualization, and augmented reality.

To be widely adopted, the interfaces of the future will need to offer alternatives, make it easier to learn commands, and minimize both physical and cognitive effort

Emmanuel Pietriga

Multimodality: The New Grammar of Interaction

The next major leap forward could come from the synergy between different modes of interaction, as combining them offers real added value. Emmanuel Pietriga explains: “Voice allows us to express an intention, gaze to point to an element, gestures to manipulate it, and the screen to display information precisely. None of these methods is incompatible with the others.”

In industrial maintenance, for example, a technician wearing augmented reality glasses could verbally query an AI assistant while inspecting a machine and receive instructions that would appear directly on the equipment in question to guide them through the task at hand.

This evolution is already evident. With the Vision Pro headset, Apple combines eye tracking, hand gestures, and voice commands to navigate a three-dimensional environment, while with Project Astra, Google is testing multimodal assistants capable of simultaneously interpreting voice, images captured by a camera, and information displayed on the screen.

The challenge, therefore, is no longer simply to increase the number of interaction modes, but to integrate them to make dialogue with machines more natural, more flexible, and better suited to each situation.

Four Scenarios for Interfaces in 2035

Depending on the role assigned to screens, the physical environment, and artificial intelligence, this transformation could take four distinct paths, which are not necessarily mutually exclusive.

Scenario 1: The screen becomes a multimodal interface

The first scenario, the one closest to current trends: the screen retains the central role it has always played, but is enhanced by new forms of gesture, voice, and eye recognition.

“In the office, an employee could start a search using their voice, select a graph with their eyes, correct data using the keyboard, and then confirm the action with a gesture. AI would ensure continuity between these different modes of interaction, without requiring a switch between applications,” explains Emmanuel Pietriga.

This hybrid approach would preserve the advantages of the graphical user interface in terms of information display, while allowing users to choose voice, touch, or the keyboard based on their preferences.

Scenario 2: The Physical Environment Becomes a Medium for Interaction

In this second trajectory, the interface gradually moves away from the traditional screen to become integrated into the objects and surfaces of our environment.

The research in which Emmanuel Pietriga is involved provides a concrete glimpse of this evolution. With TriPad, a research technique presented in 2024, his team is exploring the possibility of using an ordinary flat surface—such as a table or a wall—as a touch-sensitive medium for interacting with virtual content in augmented reality, using hand tracking. The goal is to allow users to make use of available surfaces rather than forcing them to use a single device.

In 2035, an architect could use a table to manipulate a virtual model of a building, while a technician could bring up controls on a surface near their equipment. There would no longer be a need to carry a keyboard or screen around all the time: the environment itself would become a control interface as needed.

Scenario 3: AI Agents Take Over Part of the Navigation

A third possibility: users no longer interact directly with each software program but delegate certain tasks to artificial intelligence agents.

For example, to organize a business trip, the agent would search for the most suitable modes of transportation, compare options, and make reservations, requesting approval before committing to major expenses. The traditional interface would then cede part of its role as an intermediary to a system capable of orchestrating multiple services.

For businesses, this evolution could profoundly change how they use software, as the value of these applications would then depend on their ability to communicate with and be activated by agents.

This trajectory would, however, require strong safeguards: limited permissions, traceability of operations, the ability to cancel, and human oversight to guide and monitor agents’ actions.

Scenario 4: Interfaces Become Nearly Invisible

In this fourth scenario, certain interactions become so seamless that they require almost no explicit commands. Digital systems are capable of taking context into account to offer assistance at the right moment, without systematically waiting for a request from the user.

However, even in a largely automated environment, screens would retain their full utility for comparing, understanding, and making decisions. The challenge would no longer be to make the interface disappear, but to determine when it should appear, in what form, and with what degree of autonomy.

Accessibility and Trust: The Conditions for Adoption

Accessibility will be a major criterion for the adoption of these new interfaces. Multimodality can open up new possibilities for people with visual, motor, or hearing impairments, but it also risks creating new barriers, as a gesture may be difficult to perform, or a voice command impossible to execute in a noisy environment.

The W3C’s WCAG 2.2 guidelines already emphasize the importance of preserving different input methods. “To be widely adopted, the interfaces of the future will need to offer alternatives, make it easier to learn commands, and minimize both physical and cognitive effort,” explains Emmanuel Pietriga.

Trust will also be a key factor, as these systems must guarantee data privacy, transparency in processing, and control over permissions. In Europe, the transparency requirements of Article 50 of the AI Act, effective as of August 2026, provide an initial framework for informing individuals when they interact directly with certain AI systems.

Finally, reducing the number of manual steps does not necessarily reduce the cognitive load. Overly proactive assistants could increase interruptions, suggest unsolicited actions, and make it difficult to understand automated decisions. By 2035, the true revolution in interfaces may lie in our ability to choose when, how, and why we interact with machines.

Glossary

Multimodal interface: A system that allows interaction using multiple modalities, such as voice, gesture, gaze, or touch.

Contextual interface: An interface that adapts its responses to the situation, activity, or user preferences.

Ambient interface: An interaction device integrated into the everyday environment, not always requiring a dedicated screen.

AI agent: An artificial intelligence system capable of organizing and executing multiple actions to achieve a goal, within the limits of the permissions granted to it.

We quote them

  • Emmanuel Pietriga

    A French computer science researcher, he is a research director at Inria, a French institute specializing in digital sciences. He leads the ILDA research team, based at the Inria center in Saclay. He specializes in data visualization and human-computer interaction. His research aims to make complex information easier to understand and manipulate. He also teaches data visualization and web programming at the École Polytechnique. Through his work, he contributes to the development of new technologies for exploring and analyzing large amounts of data.

  • W3C

    The World Wide Web Consortium is a standards organization that works to ensure the compatibility of web technologies.

FAQ

  • Their disappearance seems unlikely. They are expected to coexist with voice, gesture, and contextual interactions, particularly when visualization and precision remain essential.

  • It could enable users to directly express an intention, personalize interactions, and automate certain tasks without having to manually navigate between multiple applications.

  • The collection of personal data, misinterpretations, loss of control over autonomous agents, accessibility issues, and cognitive overload are major challenges.