December 08, 2017      

This week, the IBM Watson AI XPRIZE announced the 59 teams advancing in its “four-year global competition to develop and demonstrate how humans can collaborate with powerful artificial intelligence technologies to tackle some of the world’s greatest challenges.”

Initially, 147 teams entered the competition. An independent panel of judges narrowed the field based on “the standards they set in their plans, as well as the performance and scalability of their AI applications, with the heaviest weight placed on the potential to achieve exponential impact.”

The AI XPRIZE organizers also announced the top 10 teams, as the judges currently see them. Today, the judges named the top two teams so far.

“We’re very excited, and this is just the first year of this competition,” said Amir Banifatemi, AI lead at XPRIZE. “Fifty nine get to see what the judges evaluated as the top 10 in terms of progress. That does not mean that the 49 others don’t have something amazing.”

The contest’s domains include the following six categories:

  • Civil society
  • Energy and resources
  • Health and wellness
  • Learning and human potential
  • Shelter and infrastructure
  • Space and new frontiers

“The teams are working on meaningful areas,” he told Robotics Business Review. “These qualify as grand challenges for the world. We’re happy to see so many teams respond positively.”

The Grand Prize will be $3 million, with a $1 million Second Place prize and a $500,000 Third Place prize.

AI XPRIZE top 10, 2 are snapshots

“The ranking could change; certain things could develop,” Banifatemi said. “The judges are taking a snapshot of where people are now. It’s like different rounds in sports.”

IBM Watson AI Xprize logo

“This is an opportunity to explain that the teams are working on very hard problems. They need time to define them, specify data sets, and to sometimes add more technology expertise,” he added. “They may be looking for team members with field-specific knowledge or sponsors.”

The global competition started with teams from 23 countries. AI XPRIZE’s top 10 teams reflect a range of regions and interests.

The top two teams named today are Amiko AI and aifred health (both shown above). First-place Amiko is based in London and has offices in Milan. Its Respiro platform is intended to improve respiratory care by combining sensors, mobile apps, cloud-based management tools, and data analytics.

The second-place team, Montreal-based aifred health, is working to apply deep learning to personalized treatment of depression through tailored patient interaction and clinical data.

“We’re not at the finish line of yearly milestones,” explained Banifatemi. “We’re very cognizant of that. The top 10 and top two teams of this year are a way to recognize effort. Next year’s top 10 or top two might be different. There’s a lot of pressure on the top 10, but we still have two and a half more years to go.”

AI XPRIZE milestones and awards

The IBM Watson AI XPRIZE, which began last year, will also award $500,000 milestone prizes to the top 10 teams in fall 2018 and 2019.

“It’s important to share with the public what the AI XPRIZE teams are working on,” he said. “In April 2020, we’ll have the final three on the TED stage, and the judges and the public will finalize the winner.”

“Competitors must build an AI application that has a legitimate representation of technology, a high-quality AI, demonstrated human-AI interactivity, and a prediction of its impact on society,” Banifatemi said. “Every one of the final three will be a winner and will answer questions.”

“The public can’t evaluate technology, but it can voice its opinion on the impact,” he said. “The public will help decide first, second, and third.”

Judges have both AI and domain knowledge

“Since XPRIZE doesn’t judge teams, the judges have been selected by our advisory board over a year,” said Banifatemi. “As we saw the list of what the teams are working on, we realized we need judges with AI as well as domain expertise.”

The AI XPRIZE judges are independent of IBM Watson.

The AI XPRIZE is independent of IBM Watson, and its judges were selected by an advisory board.

“Judges have expertise in AI technologies such as machine learning, natural language processing, etc.,” he explained. “They understand the frontiers of AI and help make sure that AI is being applied properly.”

“A number of AI XPRIZE judges also have experience in disciplines such as psychology, education, climatology, healthcare, advocacy, and space exploration,” Banifatemi added. “Their feedback and knowledge are useful.”

“The combination of AI and field knowledge helps understand [automation’s] potential impact,” he said. “They’re currently evaluating first-year candidates to make sure their proposals are novel but not too early or too late.”

“We have three to four judges per team; nobody knew the first-round results until just a few days ago,” said Banifatemi. “They’re really trying to evaluate teams’ ability to develop AI applications up to industry standards and to improve human-machine collaboration.”

Focusing on the positive

Much media coverage has been devoted to warnings by Elon Musk and others about the risks posed by automation to jobs and even lives. However, the goal of the IBM Watson AI XPRIZE is to develop applications for machine learning that are commercially viable and benefit society.

“There are multiple conversations about AI and risk; we’re focusing on the positive aspects,” Banifatemi said. “As with any technology, with general AI, there’s a fear of it falling into the wrong hands.”

“Our job is to identify use cases where AI can be beneficial. We’ll evaluate in what ways competing AI solutions are safe, secure, and comply with ethics rules,” he noted. “We work with governments, NGOs [nongovernmental organizations], and the Partnership on AI to Benefit People and Society, as well as the IEEE, which has committees on the safety of AI.”

“In a sensational way, AI is portrayed as bad for humanity,” Banifatemi said. “We’re focused on what it can actually do. A larger conversation still has to happen.”

The idea of openness extends to the platforms used in the AI XPRIXE. IBM Watson is the sponsor, but it doesn’t require teams to use its technology.

“IBM Watson is available, but it’s up to each team what AI framework it uses,” said Banifatemi.

Categories of choice, social benefit

The AI XPRIZE stated general areas of focus, including healthcare and education, but there were proposals for other topics.

New teams can still apply for the AI XPRIZE.

New teams can still join the IBM Watson AI XPRIZE competition.

“We had [projects in] a number of other domains — some on climate, the environment, wildlife preservation, incentives for gender equality,” Banifatemi recalled. “They may not be as obvious as use cases for the public, and [they] probably need more work next year. More AI developments can be showcased.”

“In general, if we look at grand challenges — the same as for the U.N., big foundations or associations — the foundational domains that AI is addressing are in basic needs,” he explained. “We’re looking for a better understanding of how to provide more universal access to food, water, and energy.”

“A second category is how can we improve the quality of life?” said Banifatemi. “This includes education, medical diagnostics, mitigating climate change.”

“The third overarching category is exploration,” he said. “This includes exploration of space, biology, connections of the human brain and machines. There’s no excuse not to tackle basic human needs.”

‘Wild Card’ rounds

Unlike other competitions, teams can still join the IBM Watson AI XPRIZE within the coming year, which reflects the continuing evolution of artificial intelligence.

“There’s an opportunity for teams that haven’t participated yet to join,” said Banifatemi. “More than 100 teams were registered to compete in time. For reference, we had more than 10,000 applicants initially.”

“There’s new knowledge about things like reinforced learning and the capabilities of AI, as well as new resources and data sets,” he said. “GPUs [graphics processing units] are getting more powerful, which opens new doors to certain types of problem solving. We don’t know yet where we’re going. It’s natural for us to leave that opportunity open.”

“New teams need to register and demonstrate use cases, and the judges will evaluate them compared with the top 10 teams,” he said. “We’re still seeing how many will come forward by Dec. 20 for the first round. We’re still accepting teams, and we’re as eager as you are to see who they’ll be.”

There will be a second Wild Card round in late 2018.

More on Robotics and AI R&D:

Getting local support, AI ‘superpowers’

Another option for people interested in the IBM Watson AI XPRIZE is to support or join existing teams.

“We want the existing teams to get all the support they need,” said Banifatemi. “There’s no limit to their connections to experts and industries. Community support, funding, and access to more data is extremely helpful.”

“This could help teams go further, make people aware of what they’re working on, and spark conversations about what AI can and can not do,” he said. “Teams are not only made up of technologists. Nontechnical people can be aware of needs, and you don’t need just technical people to solve a problem.”

“AI is a tool set for granting people superpowers,” Banifatemi said. “We need teams that are multidisciplinary to solve these hard problems.”