Generative AI Will Transform Virtual Meetings

Your company’s video conferencing system probably uses rudimentary AI to record meetings and create transcripts. But over the next few months, thanks to generative AI, your software will gain capabilities that will completely upend how online meetings are conducted. Some of the newest features already being tested include automatically generating meeting minutes, assigning action items, creating Q&A meeting summaries, and more. 

Before vendors make these solutions generally available to clients, it’s critical that companies understand the essential factors that contribute to effective meetings. This article is intended to help anyone who regularly uses videoconferencing systems understand which features will be available soon, and how those features will impact your video calls. As the founder of a technology company that offers software engineering services to companies looking to implement next-gen technologies such as AI and machine learning, I believe the combination of generative AI and videoconferencing software will forever shape how (and how effectively) we meet. But these features will only benefit your company if you use them in ways that address common videoconferencing and in-person meeting limitations.

I have identified three key factors as the pillars of online meeting effectiveness: meaningful engagement, actionable outcomes, and fostering diversity and inclusiveness. Using these pillars, I can envision a set of possible AI capabilities that can be added to videoconferencing systems, some of which are already being tested, and some of which I anticipate will be available based on my understanding of generative AI’s capabilities and conversations with clients and partners. 

To identify and prioritize such AI-powered use cases for videoconferencing systems, I started by asking several critical questions:

• Which meeting strategies and practices are effective in the real world? How can we replicate them using generative AI?
• How can we address and overcome common challenges and barriers associated with these key factors?
• What can we learn from research from related disciplines such as team effectiveness, organizational culture, and human-computer interaction to improve meeting effectiveness?
• How can AI-driven insights and analytics be used as reinforcement for learning and improvement in meetings?

4 Characteristics of a Successful Meeting

Four characteristics of online meetings significantly impact meeting effectiveness: personalized content tailored to learning styles, purpose alignment, bias detection and prevention, and promoting active learning. By focusing on these essential characteristics, your company will be poised to deliver a significant leap in meeting effectiveness.

Below is a brief overview of each capability and how AI can drive a change in their execution.

Personalized content tailored to learning styles

We all learn in different ways. Visual learners benefit from visually appealing materials like infographics and videos. Auditory learners prefer personalized audio content and discussion-oriented sessions. Historically, it was cost-prohibitive to personalize content. But given the leap in generative AI capabilities, it is now possible to generate multiple versions of the same core content. Advanced generative AI systems will soon be able to tailor content (prep, during meeting, and follow up) to the unique learning style of each participant, boosting engagement and action orientation. Zoom is already testing a feature that allows users to chat with a generative AI “companion” during and after meetings to help users better understand the content presented.

Purpose alignment

Meetings are not one-size-fits-all. Information-sharing meetings require a different structure than brainstorming sessions. Coordination meetings have a different rhythm than monthly business reviews. Generative AI makes it possible for a bot to replicate the expertise of facilitators and guide every team through a structured process. More importantly, they feature capabilities that facilitate coaching teams and leaders.

Bias detection and prevention

Bias can manifest in subtle ways, such as men with louder voices interrupting women with softer voices during video calls. As digital meetings become increasingly ubiquitous, AI can serve as a vigilant listening post. The intention is not to intrude. But bias must be interrupted in real time. Generative AI is being tested as a way to help HR design real-time interventions to interrupt bias. Additionally, personalized recommendations will assist teams and leaders with feedback and improvement plans. HR can gain a broader understanding of the progress made against diversity and inclusion goals. It can identify potential hot spots before they escalate into significant issues.

Collective and individual learning promotion

Meetings serve as valuable ground to advance individual and collective learning. Meetings help us gain insights into our personal capabilities in areas such as communication, conflict resolution, and teamwork. Meetings help teams improve trust, accountability, and coordination. To optimize the learning experience, advanced video conferencing solutions must now be designed to foster active engagement, gain insights into team dynamics, and deliver real-time feedback during meetings. These features will enable individuals and teams to broaden their knowledge, foster idea exchange, and attain greater levels of learning and innovation during meetings. Vendors like Cisco and Zoom are already testing and building these features into their tools. Companies that purchase these systems must be mindful of how the tools are built, how they will be used, and which advanced features are necessary for the best possible collaboration experience for employees.

12 Generative AI Use Cases for Meeting Effectiveness

Based on my research into online video systems, online meetings, and generative AI, I have prioritized 12 emerging and proposed use cases of generative AI for videoconferencing. Some of the features are being tested by videoconferencing companies. Others are capabilities that are common in other generative AI systems and will likely be available for videoconferencing within the next several months and years.

Use cases for improving learning-style personalization

1.   Four narratives: This use case generates multiple narratives from one piece of core content. Each narrative caters to a specific learning style. One of my company’s partners is working with a management consultant whose core content is stored in a knowledge graph. Using the knowledge graph as base content, their system can generate three different narratives: a causality tree for visual learners, a set of hierarchical notes for read/write learners, and an audio version for the auditory learners.

2.   Simulcast multiple narratives: This proposed use case is targeted toward meetings that have a significant number of participants with diverse learning styles (e.g., all-hands meetings). Participants are automatically sent to breakout rooms aligned with their learning preferences. The core presentation is simulcast in multiple narratives using AI. For example: For the read/write learner, the system would generate additional written explanations that participants can refer to during or after the presentation. Likewise, for the visual learner, the system would share visual content like presentation roadmaps, infographics, color coding, etc. For kinesthetic learners, the system would make interactive exercises available in the sidebar.

3.   Diversity alerts: This proposed use case is focused on ensuring content and facilitation align with audience diversity. Visual, auditory, and kinesthetic learners have historically experienced challenges in engaging with meeting content (which tends to favor read/write learners). Generative AI will monitor meetings to ensure that the content and facilitation match the diversity of the audience, similar to how aforementioned videoconferencing systems are learning to adjust content to stop bias. It will provide real-time feedback if it detects misalignment. This feature will be extremely valuable in complex projects where cross-functional alignment is crucial. Lawyers are typically read/write learners, engineers are kinesthetic, and marketing people are visual. Adapting meetings to multiple learning styles will enable greater cross-functional alignment.

Use cases to align meetings to purpose

4.   Experts-on-demand: This proposed use case will enable teams to discover and utilize experts and frameworks for different meeting types through a marketplace. Each meeting type (decision-making, innovating, coordinating, etc.) has its unique structure, agenda, and facilitation playbook. Typically, teams commission expert facilitators to access this expertise. The generative AI-powered meeting systems will make it possible for a team to adopt such frameworks without an in-person expert facilitator. Organizations can license the best available framework for each meeting type and make it available to the teams via the meeting software.

5.   Real-time game board: This use case provides teams with real-time feedback during meetings by analyzing the live stream of meeting video and audio. Advanced video-conferencing systems will soon have the ability to parse a live stream of meeting video and audio and dynamically score the meeting on attributes of engagement, learning, and team environment. These metrics along with benchmarks will be shared with the team in real time, as well as for postmortems, and they can be used to encourage meeting leaders to improve future calls.

6.   Standout moments: This use case allows generative AI to automatically detect and capture standout moments from audio and video feeds of meetings. One feature video conferencing vendors are currently experimenting with is the ability to auto-detect the best moments of a meeting and make it available for subsequent distribution and discovery. Think of it as the equivalent of video highlights of a basketball game. Standout moments can be a great tool for team reflection as well as for sharing best practices.

Use cases to detect and prevent bias

7.   Bias interruption: As we discussed in a previous section, this proposed use case focuses on leveraging generative AI to detect and address biases in real time during meetings. Once an incident is detected, the software can be trained to respond based on the severity of the bias. It can either raise a warning in the chat window or, if necessary, interrupt the meeting altogether. Additionally, the systems can be further trained to facilitate remediation strategies, such as perspective-taking and stereotype-countering. By proactively addressing biases, this use case aims to create a more inclusive and equitable meeting environment.

8.   Self-awareness: This proposed use case focuses on providing leaders with comprehensive reports to increase their awareness of potential biases. These reports use direct quotes from the meeting or video clips to illustrate specific instances. The goal is not to shame or criticize but to facilitate reflection and self-correction. By offering insights into biases that may have unintentionally surfaced, leaders can engage in meaningful self-reflection as well as monitor their progress over time. This use case aims to empower leaders to proactively address their biases and foster a more inclusive leadership approach.

9.   Advance warning system: This use case aims to provide HR departments with early detection and escalation reports on bias-related incidents. The HR department cannot attend all meetings. By analyzing meeting data, including spoken words, and observed behaviors, generative AI can identify potential hot spots where biases may be present. Although generative AI can produce analytics tp help you run a better meeting, no company offers escalation reports, which can serve as an early warning signal, alerting HR to areas that require attention and intervention. This use case empowers HR to take prompt action and ensure a more inclusive and respectful work environment.

Use cases to promote collective and individual learning

10.  Real-time discovery: This use case showcases the power of AI in enhancing meeting conversations through dynamic information retrieval. Meeting software powered by generative AI actively listens to dialogue and proactively provides relevant information in the chat window. For instance, one source of confusion in meetings is how people use certain key terms like capacity, productivity, etc. Meeting software powered by generative AI can retrieve and present the organization’s standardized definitions, reducing confusion and ensuring clarity. Moreover, it can intelligently pull in contextual resources, such as video clips of the CEO’s address, to enrich discussions and provide valuable insights. This proposed use case empowers participants with real-time, context-specific information, fostering more productive and efficient meetings.

11.  Personal coach: This proposed use case is about supporting individual development goals using observed behaviors in meetings. Personally, I use meetings as a barometer to gauge progress on my personal development goals. How am I doing in handling difficult conversations? Am I able to display empathy even when I do not feel understood? Participants would be able to assess their progress in areas such as handling difficult conversations and displaying empathy. The system would provide tailored recommendations based on individual development goals and learning styles.

12. Team health barometer: This proposed use case highlights the transformative impact of AI in measuring and proactively addressing team dynamics. Although videoconferencing analytics exist, most focus on observing presenters, not the people to whom they’re presenting. This will change soon. They say people come to work for your employment brand and leave due to the team environment. It is quite daunting for any HR function to measure the health of each team and intervene proactively. For larger or project-based organizations, the complexity of managing team dynamics becomes even more challenging. Meetings are a great source of insights into team health. Using audio and video analytics, advanced generative AI systems can serve as a reliable source of input to the team and HR, allowing proactive interventions to maintain a healthy team environment.

Generative AI will forever change the way meetings are conducted. In the near future, meetings will offer personalized content and purpose-driven expertise, while also serving as guardians against bias and promoting active learning. As AI continues to advance, every meeting holds the promise of being productive, efficient, and influential, unlocking infinite possibilities for teams and organizations.

Written by:  Dash Bibhudatta, author, for the Harvard Business Review.