Recently we’ve seen an uptick in the number of bots online, which vary from chatbots for customer service to spambots on social media to content-editing bots in online communities. Though an ecosystem of bots is unfolding, our knowledge and understanding of how bots interact with one another is limited. Because bots don’t have emotions, you’d think their interactions would be relatively uneventful. However, they have the capacity to be quite social. This begs the question — what affects bot-to-bot interactions and how can developers design bots that have complex interactions without interference?
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Wikipedia’s bots recently made news because instead of editing articles on the website, they were fighting silent, tiny battles where they contradicted one another — and nobody noticed. Researchers at the Oxford Internet Institute discovered that the bots spent years doing and undoing vandalism, flagging copyright violations and more. This presents a problem not just for Wikipedia, but for all software that uses bots. Understanding the impact of bot-to-bot interaction is crucial for providing dependable bot services.
What’s a bot?
So, let’s back up for a second. We know there are chatbots, spambots and editing bots that Wikipedia uses, but what really is a bot? Operating in the background of a user’s life, a bot enables microservices that integrate deep-learning algorithms and the benefits of artificial intelligence. Bots are like a small computer program that listens to the real-time data provided by your devices. By listening in, bots are trying to figure out how to effectively understand and use that data in order to learn, react and communicate with you.
The potential for bots extends far beyond simple messaging bots. They play a huge role in the next wave of consumer solutions because they eliminate the need for a screen. Intelligence is beginning to surround us in everyday objects that are connected to the cloud, like your coffee maker or thermostat, and it doesn’t require a screen. We will know that we’ve reached peak bot potential when they deliver services that dive deep into a person’s life, rather than mimic a simple screen conversation. The bot interface is a spoken conversation. While connected outcomes today are still driven by screen interactions, the success of Siri, Alexa and other voice services prove the future is closer than many can imagine.
Pushing IoT forward with bots
True innovation in IoT can’t happen until companies apply ambient computing to smart devices — this can happen through bots. Ambient computing transforms things into intelligent devices by proactively learning patterns and influencing outcomes for a specific set of people and devices. Bots take advantage of ambient computing by learning the habits of people and automatically adjusting devices to meet their needs.
Bots have the capability to enable services that manufacturers may have once thought were impossible. Using ambient computing, bots that are specifically assigned to home security can learn to automatically disarm your security system when you arrive home, or enable settings that are designed specifically for security while you are home. A bot designed for connected light bulbs can learn your behavior and automatically adjust lighting to give the impression you’re home when you’re away. Bots learn and understand your behavior and effectively deliver the massive potential of ambient computing. Using bots, manufacturers can integrate additional useful features in their own products. By connecting their own devices into an entire ecosystem of devices and services that other developers incorporate, manufacturers can make their product entirely more useful than the product would have been otherwise.
While bots may seem very useful, they are not without flaws and currently face potential problems. Bots can disagree with one another, which can be difficult to measure and can result in unpredictable and inefficient behavior. Without a set language, rules and coordination, there would not be a way to ensure that bots can do their job.
What needs to happen with bots now?
To prevent problems from happening like we saw with Wikipedia, IoT companies need to create technology that enables bots to talk privately with each other, without exposing information about the user to the outside world. This would effectively keep them from fighting and allow them to do more communicating, coordinating and reasoning on behalf of the user in order to learn how to run their home.
It’s not enough for a developer to create a bot to act on their behalf. Bots need to go beyond acting on a user’s behalf, and instead be able to reason on behalf of the user. We’re seeing the emergence of bots communicating through examples like Clara, a virtual assistant bot that schedules meetings by using natural language to communicate directly with other peoples’ bots. Today in the world of artificial intelligence and IoT, People Power’s technology enables multiple virtual assistants that learn how you want to run the home. So it’s crucial to determine how bots are communicating — is it a natural language or machine language? Similar to how many IoT players are trying to get devices to communicate with each other today, we need an AllJoyn or an Open Interconnect Alliance for bots to talk to each other. This is an area that is ripe for research and exploration, and IoT companies must get ahead of it before bot communication protocols experience the same fragmentation as IoT device protocols.
Companies should use bots that are under control of the end user, as opposed to Wikipedia’s bots that were not under any kind of central control. In the IoT world, the end user can choose to have a bot in their account or not. At a more granular level, the user should be able to choose to give the bot permission to access a device or not. In addition, bots should have the ability to communicate with each other and coordinate activities by flocking together. This will be a huge step forward as one bot may discover you have a TV connected to a smart plug, and a different bot may recognize that you’ve gone to bed. So the two can work together to make sure your TV is turned off when you’ve gone to bed.
To avoid future bot brawls like we saw with Wikipedia’s bots, companies need to not only understand the value in bots, but also grasp how they can communicate with one another to provide the most effective, predictable and personalized experience for all users.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.