Multitasking chatbots are the new Amazon Alexa Prize challenge

In 2016, Amazon launched the Alexa Prize, an annual challenge to create AI-powered chatbots that can maintain a coherent, engaging 20-minute conversation on popular topics. A number of innovations sprang from it, including natural language understanding, neural response generation, common sense knowledge modeling, and dialogue policy technologies, leading to smoother and more engaging conversations. Inspired by this success, Amazon is today launching a new challenge aimed at encouraging developers to create experiences that can handle multistep, multimodal, and multiturn tasks.

The pandemic appears to have supercharged voice app usage, which was already on an upswing. According to a study by NPR and Edison Research, the percentage of voice-enabled device owners who use commands at least once a day rose between the beginning of 2020 and the start of April. But the uptick in adoption hasn’t correlated with an increase in voice assistants’ technological sophistication, at least in some respects. For example, voice apps remain largely incapable of replying to more than one question at a time with a single response, which can be particularly frustrating in enterprise environments where multitasking is the ordinary course of business.

The newly launched Alexa Prize challenge — the TaskBot Challenge — is intended to promote research along this direction. It focuses on improving multimodal experiences, seeking to spur the development of “immersive” and “intelligent” task completion workflows. According to Amazon, the goal is to foster chatbots that can get users from point A to point B whether they’re cooking or writing a report, or even completing DYI projects around the house.

As with the initial Alexa Prize challenge, ten teams of university students will be selected to compete for research grants, Alexa-enabled devices, free Amazon Web Services cloud computing services, and more. The application period will begin on March 17 and extend to April 16, and the TaskBot Challenge will run for three years, with an initial focus on the domains of cooking and home improvement.

“The challenge incorporates multimodal customer experiences, so in addition to receiving verbal instructions, customers with Echo screen devices, such as the Echo Show 10, could also be presented with step-by-step instructions, images, or diagrams that enhance task guidance,” Amazon explained in a press release. “The customer might ask Alexa how to fix a scratch on a car. The chatbot will then ask the customer more questions about their task, and then interactively provide step-by-step instructions and explanations for each step, or potentially adjust its plan based on customer input. After the interaction ends, the customer will be asked to rate how helpful that chatbot was with the task, and will have the option to provide freeform feedback to help the teams improve their chatbot.”

Better multitasking and multimodality could encourage enterprises previously reluctant to adopt voice assistants to give the technology a second look. In a recent survey of 500 IT and business decision-makers in the U.S., France, Germany, and the U.K., 28% of respondents said they were using voice technologies and 84% expect to be using them in the next year.

Voice assistants like Google Assistant, Cortana, Alexa, and Oracle Digital Assistant can analyze enterprise-specific vocabulary and derive intelligence from enterprise applications like enterprise resource management, customer relationship management, and HR systems. They’re also able to automate tasks like scheduling meetings, setting reminders, and assisting with conference call setup, as well as transcribing video and voice meetings.

Alexa for Business, which made its debut at Amazon Web Services’ re:Invent conference, allows enterprise users to schedule meeting rooms with third-party services such as Cisco, Polycom, and Zoom. They’re also able to share itinerary information, check voicemail messages, and quickly see if meeting rooms are available.

“Companies are using Alexa’s workplace and home office capabilities to increase productivity, improve the meeting room and video call experience, and roll out touch-free office environments,” Aaron Rubenson, VP of Amazon Alexa Voice Service and Alexa Skills, told VentureBeat in a previous interview. “Employees can use voice commands like ‘Alexa, join my meeting’ or ‘Alexa, I’m running late’ to join calls or email colleagues … As employees start returning to the workplace and traveling resumes, customers are also using Alexa for Business to provide a hands-free alternative for using shared devices like touch screens, lighting controls, and thermostats.”

The Alexa Prize seeded the development of some of the features available for Alexa for Business customers, including an explicit content filter and a neural response generator. Amazon claims that the neural response generator, which reviews conversation histories and finds topics to inform the generated responses, led to increased satisfaction among Alexa users. As something of a case in point, users have had nearly 10 million interactions with Alexa Prize-originated chatbots over the past three years, according to the company.