AI is revolutionising the way we communicate and collaborate in the workplace. With rising adoption, one area where its impact is particularly noticeable is in meetings.
By transcribing and summarizing conversations and interactions between on-site and remote attendees, AI tools such as Microsoft 365 Copilot and Zoom AI Companion can generate accurate, searchable records and identify action items post-meeting.
In addition, AI can operate in real-time, actively listening to the meeting and offering prompts that can inspire attendees with questions and ideas.
This brings businesses several productivity benefits. Employee resources are no-longer wasted on notetaking. Attendees can direct their attention and brainpower to the actual purpose of the meeting. Moreover, AI tools can also be used to capture impromptu, in-room brainstorming sessions to provide context and notes, in addition to what would usually only be an unannotated photo of a whiteboard.
Indeed, such is the utility of AI in meetings that a recent Techtelligence/UC Today survey on AI in UC showed 55.2 per cent of respondents use AI for audio transcriptions. Additionally, 43.9 per cent surveyed said they use AI for smart meeting summarisation, condensing lengthy discussions into concise, actionable takeaways and making personalised follow-ups more focused and efficient.
Such outcomes speak for themselves. Respondents of the same survey rated AI's impact on workforce productivity at an average of 3.7 out of 5 (with 5 being a high level of impact), and 64 per cent rated it at 4 or 5.
However, meeting AI tools currently face some challenges in reaching their full potential, particularly in the context of flexible workspaces and hybrid collaboration environments. Their effectiveness hinges on successfully processing quality input from the physical world, with clear voice capture being a critical, enabling factor.
Good Enough Audio Isn’t Good Enough Any More
High-quality audio is a cornerstone of effective hybrid collaboration. AI-driven tools rely on the intelligibility of the audio input to accurately do things like transcription, which forms the base of many other functions, like summarisation, personalised notetaking and follow-ups.
While UC platform providers like Microsoft Teams and Zoom are leading the market in terms of AI development, they can only apply their innovations to the audio “data” captured by the room systems. Bad data in means bad information out.
Audio quality that was deemed “good enough” in pre-AI times is no longer doing the job. Organisations that are serious about realising productivity gains from implementation of AI tools must equally ensure that their meeting room equipment captures high-quality audio so that AI models can do what they need to effectively.
The Challenges of Clear Voice Capture in AI-Powered Meetings
Imagine a scenario where you have an important meeting with your whole team, one that sets the direction for a critical project over the following months. Of all the attendees present in the room, some are seated closer to the microphone(s) and some are further away; others speak quietly and some lean back in their seats or may even move around as they speak. Layer on top a reverberant or acoustically challenging room, and you have a mixture of audio issues to contend with.
Inconsistent audio pickup and lack of clear voice capture not only make it difficult for people on the far-end of the call to hear clearly and engage meaningfully, but when the meeting finishes, your AI agent has not accurately captured the conversations either, rendering meeting summaries useless.
Equally, if missing or incorrect transcription goes unnoticed, then employees may proceed with actionables that don’t consider the complete and accurate picture of what was said in the meeting.
Backtracking a chain of decision-making also becomes a guessing game. Ambiguous and unclear voice capture means AI has a tougher time in performing talker identification and attribution, making it impossible to keep track of who made what decision in which meeting.
