The promised productivity gains from artificial intelligence and machine learning seem alluring, especially when used with collaboration technologies. Since collaboration tech can involve people in different locations and from different organizations, IT teams should start assessing AI’s impact and establishing requirements and best practice guidelines for how AI will be adopted and adapted to collaborative workflows.
We can already see AI or algorithmic data processing and predictive analytics at work in many collaboration applications. Our email and calendaring systems now proactively sort and prioritize messages and include smart reply and type-ahead features to speed message composition. Voice recognition and bots help us compose messages and interact with systems of record. The future holds the potential for much more, however.
Automated and predictive decisioning as well as robotic process automation can drive big productivity gains for companies, particularly when teams need information from multiple sources to make informed decisions to set strategies and tactics.
For example, a sales team needs to create a presentation to deliver to a client. In the future, predictive analytics may help the team members by recommending content, themes and presentation order. This technology has the potential to guide the team in building the best possible pitch. In the same way that users can edit and comment in real-time in Google Docs and Office 365, algorithms have to potential to tap data from CRM systems to guide content creation. By looking at data from a prospect’s profile and peer organizations, as well as the content that contributed to successful sales in that peer group, the application could intelligently guide the team to the best presentation possible.
Taking this example a step further, robotic process automation can be used to analyze public information about a company--such as news stories, data from financial filings and earnings call transcripts--to identify issues with a customer or prospect and proactively alert an account team with recommended actions. The process automation system could then integrate with collaboration tools such as email and calendar applications and CRM systems to proactively create a set of individual and shared tasks to improve customer engagement.
Technology providers--particularly in the video, web and voice conferencing market--have begun to deliver a vision of the future in which AI can drive productivity gains through automated meeting join and call control, identification of participants and transcription of audio. While all of these capabilities can deliver big benefits for users, companies need to consider how well they work and what, if any, impact they may have on existing policies and procedures organizations have in place.
Cisco, for example, has announced a set of features for Webex that can help automate and enhance meeting experiences. The Webex Assistant includes capabilities designed to ask if a meeting organizer wants to join a web meeting when he or she walks into a meeting room. This capability, called Proactive Join, uses Cisco’s intelligent proximity technology to determine user location from a mobile device and correlate it with the user’s calendar. Via speech recognition, the user can confirm or defer joining that scheduled meeting.
Other vendors will certainly offer similar solutions, so large organizations with multiple sites and technologies have to consider how rolling out the technology to select meeting rooms or locations could impact user behavior. If users become conditioned to using the technology, how will they react when it isn’t available?
Participant recognition in video conferencing also has some wider implications. Vendors are marketing the tools as able to search internal and external sources and use facial recognition to identify meeting participants on screen. While certainly a helpful convenience to identify participants, some people may find this idea intrusive. Furthermore, some facial recognition technology has been demonstrated to misidentify people based on gender and skin color. If the solution isn’t 100% accurate, this can cause brand damage for both employees and third-party meeting participants.
Services that transcribe speech in conference calls to text meeting notes have also come to market either as a built-in capability of a conferencing solution or as a third-party add-on. Just about any team could make use of this capability, especially as speech recognition becomes more accurate. As tools develop that automate the task of taking action items discussed and assigned during a meeting, they will be able to automatically populate various collaboration and other enterprise systems of record with tasks, timelines and data updates.
IT organizations need to work with their legal departments to establish guidelines and policies related to the use of these services. The legal consent to be recorded is the critical element companies have to understand and establish policies around, so that users adopting these services do so appropriately.
Furthermore, organizations need to assess two risk factors for these transcripts.
First, they need to decide if and how these transcripts fit into data retention policies. Questions IT managers need to ask include: Do both text and speech need to be stored and for how long? Where will these transcripts and recordings reside, and how can they be made accessible for e-discovery?
Second, IT departments need to know what risks may be associated with the way a transcription service provider trains its speech recognition algorithm. Recently we’ve learned that vendors with speech recognition technology have been using humans to listen to recordings to help train algorithms. Imagine if a board of directors meeting discussing an acquisition was part of a training exercise that involved a person listening to even a partial recording of the call.
The potential of AI and machine learning to transform collaboration and improve productivity definitely exists for organizations, and IT organizations should actively be looking at where and how the technology can be applied most usefully. At the same time, IT also has to be smart about the risks.