Nand Kishor is the Product Manager of House of Bots. After finishing his studies in computer science, he ideated & re-launched Real Estate Business Intelligence Tool, where he created one of the leading Business Intelligence Tool for property price analysis in 2012. He also writes, research and sharing knowledge about Artificial Intelligence (AI), Machine Learning (ML), Data Science, Big Data, Python Language etc... ...Full Bio
Nand Kishor is the Product Manager of House of Bots. After finishing his studies in computer science, he ideated & re-launched Real Estate Business Intelligence Tool, where he created one of the leading Business Intelligence Tool for property price analysis in 2012. He also writes, research and sharing knowledge about Artificial Intelligence (AI), Machine Learning (ML), Data Science, Big Data, Python Language etc...
3 challenges of developing bots for immersive environments
Facebook recently announced that the chatbots on Messenger platform have been having problems, and indeed have a failure rate of some 70 percent.
On the VR/AR front, Magic Leap has run into problems too, with reports of its product lagging behind expectations. To quell hostile rumors, Facebook is emphasizing menus rather than free text for chatbot interaction, and Magic Leap is pushing for a 2017 launch.
These hiccups exhibit the still-persistent challenges in computing power and energy management with VR/AR and cognitive resources and technology within AI. But most agree that these are technical barriers that will be ironed out within a few years. Following IXDA‚??s Interaction 17 conference, I wanted to dive deeper into the design and human issues with intelligent and immersive interfaces, especially along three clear themes that I see emerging.
1. Balance focus and freedom to meet expectations
One of the hiccups with the great AI hype of 2016/17 has been that people now expect AI to work out of the box like Tony Stark‚??s Jarvis. While we can already build creative AI applications for narrow areas and cover broader fields with mostly retrieval-based systems, a general AI capable of generative thought and output in an open-domain setting remains elusive.
For this reason, framing the problem and communicating it effectively is crucial. It helps if you help the audience understand the purpose of the experience, set expectations levels for both the organization and the user, and enable the development team to set metrics for evaluating success and failure.
On VentureBeat‚??s 2016 Best Bots list, productivity bot Meekan is a good example of great framing. Its task is to simply find time on people‚??s calendars and set up meetings. The Slack environment, where all users have their calendar access shared, further helps frame the use cases and help craft successful experiences.
Framing is important in the VR/AR world also. With traditional content, the creator has great control over the viewer‚??s attention through selection of subject matter, framing, and timing. VR/AR provides the viewer with 360 degrees of freedom, with the added dimensions of depth and time. This complicates things and can be distracting.
Gary Hustwit, a pioneer in immersive content, points to Katy Newton‚??s user research in VR environments that shows how increasing degrees of freedom in immersive environments can lead to decreased retention of the narrative‚??s key elements ‚?? and the reverse can happen as well. Visual and aural cues, as well as narrative techniques, are needed to direct attention effectively.
2. Build agency, not just agents, to win trust
A few people at Interaction 17, including Zachary Jean Paradis, raised the challenge of agents versus agency. As the Facebook chatbot failure statistics show, people too often feel like test puppets for subpar AIs instead of perceiving them as helpful. To fix this, companies need to design intelligent and immersive experiences with special focus on how the user exerts power and achieves desired ends.
As an example of a flawed agent, or a ‚??dumb smart thing‚?? that fails to empower users, Mr. Paradis pointed to Nest and its Auto-Away function, that has been received much criticism on places like Reddit. That type of blunder in daily functionality risks frustrating and alienating users, even if other aspects of the service were in order.
Cennydd Bowles noted that machine learning is too much like ‚??money laundering for bias‚?? and called for design thinkers to make progress with ethics, always aiming to improve the lives of users and providing them with understanding of not just what‚??s being done but how and why are these decisions being made. He also had a practical checklist to test if what you‚??re designing passes the test of applied ethics in design:
Am I treating people as ends or means?
What if everyone also did what I‚??m about to do?
Am I maximizing happiness for the greatest number?
Would I be happy if what we‚??re doing was in tomorrow‚??s paper?
Brenda Laurel has set up a useful framework for providing agency. According to her, properly designing the frequency, range, and significance of choices available to a user is key. Because immersive experiences are much like interactive stories, the arc of drama and pacing of these moments matter greatly. Overly frequent choices that carry too little ‚?? or too much ‚?? significance risk bugging or burdening the user needlessly, so it‚??s important to mix lighter and heavier choices for an optimal rhythm.
3. Design friction to drive reflection in an on-demand world
We live in an increasingly frictionless and accelerating world. You can now order a pizza with zero clicks, Amazon is gearing up for huge airships and 15-minute deliveries, and banks are planning to let machines run much of their operations. When the pace of decision making by people and computers alike picks up, purposefully thinking about the consequences of these decisions becomes increasingly important.
Experience changes you. But what happens when experiences become too smooth? According to Airbnb‚??s Steve Selzer, immediacy and the absence of friction are creating a less tolerant, less self-aware world. This is why designers of intelligent, immersive experiences need to build in meaningful friction, encouraging reflection and awareness of the actions themselves as well as their consequences.
It is also not just a matter of humanity. The harmful effects of completely frictionless experiences are known in economics as facile externalities. These include consumers spending too much, customers becoming alienated, and people and societies ending up worse off as instantly gratifying but selfish habits proliferate. More provocatively, Jared Ficklin talks about ‚??Death by 4.7 stars,‚?? pointing out that an evil AI wouldn‚??t need killer robots ‚?? it could simply recommend Jolt cola, hamburgers, and ice cream to buyers happily eating their way to oblivion.
In addition to fast conversational experiences, reflection is even more important in immersive environments, where you don‚??t so much ‚??watch‚?? or ‚??use‚?? experiences as really ‚??live‚?? through them. VR experiences are perceived by the brain as actually happening to user, so their transformative potential ‚?? toward self-development or rapture ‚?? is quite powerful.
Even with good intentions, if reflection is absent, computers nowadays can wreak significant havoc. How to provide the right amount of friction to unlock reflection but not to hamper experience is therefore critical in building a world that, in addition to doing things, also thinks about what it‚??s doing. On the flipside, technology is promoting some good behaviors too. People seem to thank chatbots surprisingly often, although there is no logical reason to do so.