What if you could determine instantly when your webinar speakers are boring the online audience and nudge them to liven things up? What if you were alerted that a virtual education event was causing confusion instead of creating an aha moment, and make sure ideas were clarified?
Those techniques are on the horizon, as new emotion-analyzing software is being applied to everything from advertising to gaming. Officials at Waltham, Massachusetts–based Affectiva, an emotion-recognition technology company, say applications to online education and virtual events aren’t far behind.
Affectiva’s Affdex software uses complex algorithms that scrutinize and identify facial expressions. Developed in part by Affectiva co-founder Rana el Kaliouby through her work at the Massachusetts Institute of Technology’s Media Lab, the emotion-sensing analytics program has a variety of possible uses. Convene recently talked to Gabi Zijderveld, Affectiva’s vice president of marketing and product strategy, about what it could bring to online programs.
How versatile is emotion-recognition technology?
The technology is in the cloud, and all we really need is a camera to record or analyze someone’s face. Our technology measures, frame by frame, how a person responds, analyzing their emotional journey. It’s pretty unobtrusive and super-scalable. For market research and advertising, we can test how consumers respond to brands and messages and digital content, such as ads. In the past, they’d do that through focus groups or surveys, which can be quite biased.
How might this be applied outside marketing and advertising?
Our technology is used to get insight and analytics. But our technology can also be licensed by others who can then add the ability to sense and analyze emotions to their own applications. For example, with gaming, it could allow the game to adapt to the emotions of the player. If someone is frustrated, maybe the game offers up a different challenge. Real-time adaptation and interaction are key in that scenario.
What about online education?
There are a few different scenarios, including with MOOCs [massive open online courses], where large numbers of people are taking a course with an instructor on the end. Or it might be a videoconference scenario done through a virtual classroom. Online, it’s difficult for instructors to get a read on their audience.
So an instructor could tell if the audience is getting bored or if they’re engaged?
Exactly. It would be a real-time attention meter or indicator of understanding and comprehension. In real time, an instructor could get simplified feedback to indicate, for example, that 80 percent of their audience is confused, or 95 percent just laughed at that joke, so keep it going.
You can get more in-depth analysis after an event has taken place: How well am I delivering this? How well is the content being received? Maybe there’s an opportunity for coaching the instructor, or maybe it’s an issue with the curriculum. The analysis after the fact would be meaningful for the educator, but also for those who developed the educational content.
How else could the information collected be applied? It seems like there’s an opportunity for personalization here.
A platform provider could communicate out to the audience that this particular session is getting a four-star rating and this is something you want to see. Training sessions that are webcast and prerecorded could be ranked or potentially personalized. If we know you really liked this instructor and this training course, others who have similar types of emotional profiles might like this as well. In the future, it offers a tremendous opportunity for personalization.
What about for the audience member being analyzed? Are there benefits for them?
If people have a better learning experience, it’s better for everyone. Dropout [from online-education programs] is a huge issue, and you don’t want people to fail to complete their learning. When you understand the emotional engagement of the learner, you can adapt content in real time. If they’re confused, the learning-management system could offer up a different path for that particular piece of content. Or maybe someone is bored and the system will pick up the pace.
If enough data is gathered in the educational space, you can see certain patterns and trending, and do predictive analytics. What if you could see a pattern in a learner’s emotional behavior and say they are showing signs of dropping out in the next five minutes? Then you can take action. You can start doing predictive analytics to raise flags. Instead of taking corrective actions after the fact, you start noticing much sooner that there’s a problem and you can intervene. This could potentially get super-interesting.