Voice AI to Replace Typing at Work by 2028, Study Finds

Corporate keyboards face obsolescence within two years as voice-driven artificial intelligence prepares to become the default workplace input method, according to a joint forecast by the London School of Economics and audiovisual technology firm Jabra.

The projection targets 2028 for a fundamental shift in how professionals generate documents, emails, and code. Rather than manually striking keys, workers will dictate instructions to generative AI systems that instantly structure spoken thoughts into polished text. The keyboard transitions from a primary creation tool to an editing interface used only for final refinements.

This anticipated workflow mirrors the architecture already deployed in consumer smartphones through tools like Apple Intelligence and Android’s Gemini integration, where on-device speech recognition drafts messages before transmission. Enterprise software developers are racing to replicate this frictionless pipeline inside proprietary environments. Microsoft has steadily expanded voice-to-text capabilities within Copilot across Word and Outlook, while Slack and Microsoft Teams have integrated asynchronous audio messaging features that increasingly rely on automated transcription.

Jabra global head of brand communications Paul Sephton frames the transition as an inevitability driven by cognitive alignment. Speaking processes information faster and more iteratively than typing, allowing professionals to capture ideas outside traditional desk environments. For parents managing childcare or field workers operating away from desks, voice-first interfaces eliminate the physical bottleneck of a keyboard. By the time Generation Alpha—defined as those born from 2010 onward—enters the labor market around 2030, spoken interaction with enterprise software is expected to be fully normalized.

Academic researchers caution that the mechanics of sending work differ drastically from receiving it. Fabrice Cavarretta, an associate professor of management at ESSEC Business School, argues that raw audio will never replace text on the recipient’s end. Reading speed significantly outpaces listening speed, and searching a written document for specific keywords remains vastly superior to scrubbing through an audio timeline. Cavarretta anticipates organizations will adopt a hybrid model: employees speak their inputs, but AI systematically transcribes and formats those inputs into standard text before distribution.

That transcription requirement exposes severe corporate governance risks. Bertrand Audrin, an assistant professor in human resource management and organizational behavior at EHL Hospitality Business School, warns that untranscribed voice notes erode institutional accountability. Unlike written emails, which create immutable audit trails required for regulatory compliance and legal discovery, ephemeral audio files complicate efforts to trace decisions months after they occur. Financial services, healthcare, and publicly traded companies operating under strict record-retention mandates cannot legally substitute casual voice memos for documented correspondence without automated, searchable archiving.

Beyond compliance, the ease of recording audio introduces new productivity threats. Lowering the barrier to communication often generates volume rather than clarity. An influx of poorly structured voice notes forces recipients to spend additional time parsing tangents, filler words, and ambiguous directives. Non-native speakers and individuals with speech impediments face compounded friction when forced to rely on vocal articulation instead of deliberate written composition.

Major enterprise platforms are currently attempting to resolve these structural conflicts by embedding real-time transcription directly into audio workflows. In early 2026, both Zoom and Microsoft expanded native AI summarization features that automatically convert meeting recordings and asynchronous voice clips into searchable, indexed text blocks, signaling that the industry is actively building the exact transcription infrastructure academics identify as mandatory for a voice-first workplace.