AI shapes the future of web conferencing platforms

In the last couple of years, videoconferencing platforms have become very popular and have captured the attention of researchers and the public, particularly due to the Covid-19 pandemic. Immediately after the pandemic started in the spring of 2020, demand for video conferencing apps has grown exponentially. This demand, and the emergence of new needs, have forced manufacturers to adapt to the new context by improving the services offered and adding new features to existing applications. Many new video conferencing applications have also emerged with this demand. This paper presents a series of statistics on the evolution during the pandemic and the current status of the main video conferencing systems. The different ways in which these systems have integrated artificial intelligence technologies to address different identified problems and user needs are also presented.


Introduction
Collaborative decision-making is commonly associated with meetings. The early GSS (Group Support Systems) were related to the usage of EMS (electronic meeting systems) to support the set of decision-making activities or parts of it. They were meant for rather numerous and properly organized groups (made up of ten to hundreds of people) to improve the effectiveness of the participants' work [1] [2]. The conference term is defined in the Merriam-Webster dictionary as "a meeting of two or more persons for discussing matters of common concern". It comes from the Medieval Latin term conferentia (contribution, discussion) noun, which is derivative from the

Introduction
Collaborative decision-making is commonly associated with meetings. The early GSS (Group Support Systems) were related to the usage of EMS (electronic meeting systems) to support the set of decision-making activities or parts of it. They were meant for rather numerous and properly organized groups (made up of ten to hundreds of people) to improve the effectiveness of the participants' work [1] [2]. The conference term is defined in the Merriam-Webster dictionary as "a meeting of two or more persons for discussing matters of common concern". It comes from the Medieval Latin term conferentia (contribution, discussion) noun, which is derivative from the present participle of conferre (to bring or take, convey, bring together) verb [3]. When a conference is supported by electronic means the term conferencing is used [4]. While a meeting has a goal, say a decision, the conference is meant for information sharing on a certain subject [5]. A particular class of conferences is formed by decision conferences. They address "hot issues which represent concerns of an organization that require resolution. They can be strategic issues that require work over months or years, or operational ones, including immediate crises. Decision conferences don't work very well for issues that are merely 'interesting,' or 'nice to consider,' but lacking any sense of urgency for their resolution" [6]. Collecting information in the Intelligence phase of the decision process model of H. Simon [7] or reviewing the available choice approaches to be used by the stakeholders in the possible decision situations are possible usages of decision conferencing. In this paper, the technical attributes of various conferencing solutions are reviewed and compared in the particular context caused by the Covid pandemic.

Data on the most known videoconferencing platforms
The coronavirus pandemic has created a global crisis that has forced social distancing and a shift to remote working in all areas of work [8]. Until the Covid-19 pandemic, web-based videoconferencing systems were used regularly, mainly in certain areas and by certain categories of people (e.g., managers and trainers). With the pandemic, companies, institutions, and people had to adapt to the new conditions. Thus cloud-based video conferencing apps such as Zoom, Microsoft Teams, Google Meet (Google Hangouts rebranded in April 2020 to Google Meet), and Cisco Webex, have been adopted, in all areas, for work collaboration and keeping in touch with family and friends, at a very accelerated pace, since the early months of the pandemic. The lockdown during the pandemic led to a general explosion of all video conferencing products, as many schools and businesses purchased video conferencing tools to keep people connected from home. Nowadays, video conferencing platforms are considered the most important business communications tools and a feature-rich video conferencing solution can empower the teams to be productive, making it easier for them to collaborate, problem-solve, take decisions and hold effective meetings in real-time [9].

Strengths
There are many benefits of using virtual teams and web collaboration technologies. The most important benefits are related to expenses and time-saving. Web conferencing can reduce travel and facility costs, but also, in the case of businesses, can lower project schedules, and improve decision-making time and communication [10]. For this reason, "organizations and corporations are adopting video collaboration solutions with the aim to make faster decisions and avoid high costs associated with travelling" [11].
In [10], there are presented the terms video conferencing and web conferencing, as used in the literature at that time, as referring to two different categories of online communication systems, in terms of the facilities offered. Videoconferencing systems were considered to be those systems that only offered audio-video communication, while web conferencing systems also offered other facilities such as screen sharing, chat, or polling. Over time, the terms have become synonymous in the literature, referring to the same type of systems allowing audio-video communication, regardless of the additional facilities offered.
According to [12], the best video conferencing apps: (1) offer high-quality video and audio, reliably, (2) make it straightforward to start, schedule, and join meetings, and (3) include collaboration features (at least screen sharing and chat), (4) allow for video recording, (5) make it simple to add people outside the organization. In [12], a list of the 5 best videoconferencing platforms for teams in 2022 is presented, following the testing of 30 such applications. Each of these 5 apps, in [12], is considered best for a particular case: (1) Zoom is considered the best video conferencing app for reliable, large video calls; (2) Google Meet for Google Workspace users; (3) Microsoft Teams is the best combination of team chat and video conference app; (4) Whereby (Web) is the best video conferencing app for a lightweight option and (5) [20]. All these authors identified different strengths of each of the three solutions compared to the others, depending on the context, with some of them ultimately considering the best video conferencing system: Zoom [13] [17](for education) [18](for mass audiences) [20], Google Meet [15] [18](for corporate settings) [19] or Microsoft Teams [16] (for education) [ Best features screen sharing [13] breakout rooms [14] Participant Reporting feature [15] live captioning for audio and video chats [12] [17] screen sharing options and controls [13] dial-in for free with any of paid plans [14] [17] [19] whiteboard [12] breakout rooms [14] Together Mode [14] best whiteboard [12] best track record of security and the most transparent privacy policy [14] all-in-one collaboration hub [14] [17] automatic captioning [17] Best for… In general, the authors of these comparisons agree that all these systems have similar functionality, and the choice of a particular videoconferencing system depends very much on the needs of the users. Zoom seems to be a good choice for webinars, conferences, and large groups. Microsoft Teams and Google Meet are perfect for Microsoft 365, respectively, for Google Workspace users. Depending on the purpose of the meeting, one videoconferencing solution might be preferred over another. For example, for education, each of the three platforms, and also many other similar platforms, offers enough useful options to provide a valuable and effective learning experience. The choice of a solution depends on the policy of the educational institution. Virtual decision-making meetings require more careful pre-planning and a lot of offline preparation. Tools such as Google/Microsoft Forms (e.g., to identify key issues) or Google Docs/File Sharing in Teams [23]/ Microsoft Loop (e.g., for shared work and collaboration) that help in planning and running decision-making meetings may tip the balance, in this case, towards Google Meet and Microsoft Teams.

Statistics
The emerging trend of remote working, geographically scattered business operations, and increased globalization are the major factors for the global video conferencing market size to expand at a compound annual growth rate (CAGR) of 12.5% from 2022 to 2030 [24]. In 2021 the videoconferencing market stood at USD 6.28 billion, and it is expected to reach USD 14.58 billion by 2029 [11].
According to a study conducted and published in March 2020, during the week of March 15-21, 2020, business apps recorded the highest number of downloads in the App Store and Google Play in history: 62 million downloads, with 45% more than the week before [25]. For example, Zoom Cloud Meetings, during the week of March 15-21, was downloaded 14x more than the weekly average during the first four months of 2019 in the US, 20x more in the UK, 22x more in France, 17x more in Germany, 27x more in Spain and 55x more in Italy. According to [25], similar increases were also recorded by Google Hangouts Meet and Microsoft Teams.
This growth in downloads was even higher in the second quarter of 2020. Together, Zoom, Microsoft Teams, and Google Meet had 147.4 million installs in the first quarter of 2020, growing to 479 million installs in the second quarter of 2020 [26]. According to the same source [26], the number of new installations was also high in the next four quarters, until June 2021, but their order of magnitude was in the range of 200-300 million.
In July 2022, data.ai was used to obtain a relative rank of average total time spent in the video conferencing apps or mobile websites during the May 1 -June 30, 2022, per daily active user. After selecting Google Meet as an app of interest, data.ai suggested 9 other similar apps that were used to achieve the desired ranking. The set of apps analyzed included (1) (Table 2, column 1). In the two months (Mar 1 -Apr 30) before the period under review, the top was similar, with two differences: Google Meet ranked #4 and Slack #3; JioMeet ranked #8 and Zoom for Intune ranked #7. If Skype for Business and Skype are also analyzed, they rank first and seventh respectively in the most used apps in June 2022. WebEx, Teams, and Meet are in 2nd, 3rd, and 4th place. In Romania in June 2022, Skype and Skype for Business ranked #6 and #7 respectively.
All this data shows that Zoom, Cisco WebEx, Google Meet, Skype for Business, and Microsoft Teams are among the most used videoconferencing systems.

AI in video conferencing
Recently, researchers and producers are interested in improving video conferencing systems by integrating artificial intelligence technologies [27] [28] [29]. Artificial intelligence (AI) methods, especially those based on Machine Learning (ML), can strongly contribute to creating a better user experience.
Since providing reliable, high-quality video and audio is the most important aspect of video conferencing, most of the research and developments are in this direction.
According to [30], Zoom adopted artificial intelligence at the core of its product and every aspect of the product uses some sort of Artificial Intelligence: virtual backgrounds (image segmentation), video compression (the video algorithm identifies the most relevant sections of the imagefacesand maximizes the resolution of those areas), audio processing (noise suppression, voice activity detection, speaker recognition, speech enhancement, and music detection), live transcription (Speech-to-Text, translation, summarization, and sentiment analysis), anti-fraud (to prevent unwanted abuse like pornography, telephone fraud, and spam), integration with Google's AI Assistant and Zoom Rooms (auto-tracking cameras, speaker focus, and people counting, etc.).
Google Meet uses artificial intelligence to enhance video quality: low light mode (automatically adjusts the quality of the video), autozoom (zoom in and position users squarely in-camera), data saver (helps to limit the use of data on mobile networks), noise cancellation, auto meeting notes (live captions) and video background.
Google Meet announced in May 2022 another series of AI-powered updates: portrait restore (enhance video despite bad lighting, low-quality web cameras, or bad WiFi connection); portrait light (simulate studio-quality lighting and let people adjust the lighting position and brightness), de-reverberation (improve shallow audio sounds for poor room acoustics), live sharing (sync content that's being shared in a Google Meet call and allow participants to control the media) and automated transcription (from a Meet to a Google Doc) [31] [32].
In June 2022 it was announced that Microsoft Teams is getting some AI-powered improvements aimed at boosting audio and video quality on calls: echo cancellation, de-reverberation; real-time screen optimization for video playback and brightness, and soft focus video enhancement [33]. Other useful Microsoft Teams features are the full-duplex (two-way) transmission of audio that allows users to speak and hear others at the same time and machine learning-based background noise suppression [34].
One interesting AI development for video conferencing of NVIDIA researchers is Vid2Vid Cameo, one of the deep learning models behind the NVIDIA Maxine software development kit for video conferencing, that uses generative adversarial networks (known as GANs) to synthesize realistic talking-head videos using a single 2D image of a person. After submitting a reference image (photo or avatar) before, and during the meeting, the AI model will capture each individual's real-time motion and apply it to the previously uploaded still image. Besides helping meeting attendees look their best, this AI technique also shrinks the bandwidth needed for video conferencing by up to 10x, avoiding jitter and lag [35].
Mendes et al, in a paper published in 2020 [29], present a list of eleven use cases where AI could be used in the future to solve various problems encountered in video conferencing systems: (1) sensitive/inappropriate video classification, (2) audience counting (face detection and counting the number of attendants using the same camera), (3) identity verification (face recognition to verify the identity of participants and compare them with a previously authorized list), (4) automatic layout (object detection together with automatic zoom and replacement of objects, to give a better sense of continuity among different physical spaces. Microsoft Teams has already implemented such a feature -Together Mode -that aims to improve the users' immersion experience), (5) automatic accessibility (audio description of video content for visually impaired people and recognition and translation of the spoken text into an appropriate sign language), (6) participant anonymization (use of image filters on the faces of those participants who do not want their image to appear in the meeting recording or even during the meeting), (7) video summarization, (8) user attention/engagement detection (sentiment analysis using users' face to help to classify the attention and engagement of users during the video conference), (9) live comments sentiment analysis (sentiment analysis done over users' information shared in a video conference to classify the interest in topics discussed), (10) personalized content (based on user identification and sentiment analysis), (11) virtual conference assistant.
IntellectData presents, in [36], different ways AI could transform video conferencing: (1) by tracking meeting statistics, AI could make predictions and suggest useful applications and documents be used during the meeting, or automate tasks such as rescheduling calls, booking meeting rooms based on email conversations; (2) natural language processing implementation facilitate automated transcription of meetings, translation of the conversations into different languages, suppress echo, cancel background noises, balance sound automatically, allow chatbots to manage and take over a meeting using a voice command feature; (3) computer vision improves video quality, allows active participant framing, virtual backgrounds, upscaling resolution to improve picture quality; (4) augmented reality technology can redefine and enhance communications. It is expected in the future to continue the adoption of advanced technology solutions driven by AI in videoconferencing.

Conclusions
The Covid-19 pandemic marked a milestone for the video conferencing sector. At the start of the pandemic, organizations were taken by surprise by the urgent need to use videoconferencing systems to cope with the new situation. Some of them did not require employees to use a particular platform, more so in education. Thus, each of the main platforms available has attracted a number of users. This led many organizations to adopt a multi-platform approach, where several similar tools are used within the same organization, but for different purposes and situations [17]. However, adopting and using a certain solution within an organization needs a systematic approach that takes into account various factors [37].
It is important to mention that over the last few years when the use of these videoconferencing systems has reached an unexpectedly high level, the manufacturers of these platforms have tried to respond promptly to the different needs of users by adapting existing functions and adding new ones. Many popular video conferencing platforms have been enhanced with artificial intelligence capabilities to improve both online meetings and postmeeting workflows. As presented in the paper, it is expected that artificial intelligence will be used in the future to provide a better user experience and improve the productivity and efficiency of online meetings.