We’re all used to technology changing so fast that, when it doesn’t change much for awhile, you can start to think we’ve reached the end of history. Maybe we’ve finally figured out the best way for humans to use technology to accomplish that particular task, so it’ll never change again?
It’s been starting to feel that way with meeting collaboration. We’ve been conducting meetings pretty much the exact same way since the pandemic took hold: We use a videoconferencing platform (though increasingly, people turn their video off whenever they can get away with it). We share our screens to discuss documents (though unless it’s a very focused project team, it’s rare to truly “collaborate” on a document in ways that require ideation-type applications). In between meetings, if we’re lucky enough to have such interludes in our day, we text via the collaboration app, or send emails.
So is that all there is? Is that all we need?
The vendors, of course, are betting that we need—or at least desire—features like meeting summaries and notes. Companies like Microsoft even suggest that the next wave consists of users adopting a more asynchronous approach to their meeting schedule--being able to skip meetings and review them later, and eventually even have AI-driven personal agents that gather information and take actions on their behalf.
Pretty clearly, the next generation of collaboration will be powered by AI, which opens up a host of potential problems, many of which will have to be addressed by IT--generally in concert with organizations such as HR and Legal/Compliance. Another key factor will be end users getting comfortable with AI—and that may not be so simple.
This article from ITPro Today describes some recent cases where users’ mishandling of AI-produced collaboration data caused serious problems for enterprises. In one case, the AI-driven transcription system continued recording after a call was finished, then automatically emailed the highly unflattering results to participants who, to say the least, shouldn’t have received them. In another example, enterprise leaders recorded a meeting where they discussed who would be included in upcoming layoffs, then allowed the video to be found by others.
The thing to note about such cases is that they’re not examples of AI going rogue. In most of these cases, the software did what it was configured to do—it just wasn’t configured appropriately. It seems that neither IT nor the users anticipated how the system might behave in real-world scenarios.
One refrain I’m starting to hear from IT people is that they want to get their users accustomed to AI by gradually introducing AI-powered technologies and letting users gain some comfort with them. At Enterprise Connect AI last week, Mitch Lieberman of Fidelity Investments said in a session that he doesn’t necessarily try to pitch AI transformation to those who will be using the technology. Instead, he asks them to test out limited use cases and lets them get comfortable before gradually ramping up.
In the early days of COVID, enterprises and their end users learned to develop an etiquette and set of best practices around using desktop video. IT needed to learn, for example, to configure default settings on video calls with an eye toward security, while end users got their heads around which behaviors and backgrounds were acceptable, and how to set up their workspace to optimize for video.
Now as we confront the next generation of collaboration systems, which are all about AI, everyone has some learning to do. AI may be nothing more than a tool, but IT and end users both need to make sure they know how to wield this new tool.