"The third wave" of AI and systems for state security

    There are many different hackathons and short-term contests for the practical application of AI. Two large-scale competitions stand out against their background, which pursue global goals and stretch over several years.

    The organizer of the first competition is XPRIZE, a large fund that supports innovation. The second competition is held by the United States Defense Development Agency (DARPA). The goal of some is to find solutions to global problems using AI, others want to create a “third wave” of artificial intelligence in order to improve the security of the state.

    Under the cut, we collected what is known about each of the competitions.

    TED 2016. David Kenney (IBM) announces the launch of the IBM Watson AI XPRIZE competition

    IBM Watson AI XPRIZE

    In 2016, at the Ted Talks conference, IBM and XPRIZE jointly announced a three-year team competition in the field of cognitive computing. The motto of the competition - "Artificial intelligence will solve world problems."

    The stated goal is a breakthrough in cooperation of robots and humans. The organizers believe that it is the partnership between man and machine that will help solve the problems of modern society.

    The unique feature of IBM Watson AI XPRIZE is that the competition is open, which means that the participants themselves choose their task. The organizers emphasize that there are many problems in the modern world, and with the help of artificial intelligence one can both reduce the extinction of animals and insects, and stop the sale of sexual services.

    Such a (non) formulation of the problem looks logical if we take into account the specifics of XPRIZE as a foundation that supports innovation in general - to improve the life of humanity.

    Stages of the competition

    The competition is divided into several rounds, and the participating teams annually present their designs at the World of Watson Conference.

    The interim competition of 2017 was held during the Neural Information Processing Systems conference - 10 teams took part in it, two of which received a reward. The winner was the Italian platform Amiko AI . The team developed a smart asthma inhaler and collected a $ 10,000 prize. Second place and $ 5,000 went to aifred Health - The team uses depth learning algorithms to create individual programs for treating depression.

    Among the solutions that are included in the top 10 are AI, which will help reduce poverty, as well as a system that predicts possible overdose in people with addictions.

    Participants and prizes

    Today, 62 development teams from 15 countries participate in the competition. Ahead - Second Wildcard Round, which you can sign up to December 14th - this is the last opportunity for those who want to join the competition.

    Of the teams that recently entered the competition are the Dutch OptOSS, the Israeli platform Zzapp Malaria and Mt. Cleverest.

    OptOSS develops artificial intelligence that makes a weather forecast and predicts possible natural disasters — it helps prepare for and reduce losses. The guys from Zzapp Malaria use artificial intelligence technology to monitor the spread of a dangerous disease - malaria. And the team is Mt. Cleverest is struggling with problems in the field of education - its AI system makes learning in various courses more accessible and more convenient.

    Participants of the third round will be announced in January 2019, and on September 1, teams will have to present their progress.

    The competition will end on April 1, 2020 during the TED Talks conference, where three finalists will present their decisions.

    The prize fund is $ 5 million. The winner will receive a $ 3 million Grand Prize. The runner-up team will collect $ 1 million, and the prize for third place will be $ 500,000. The remaining $ 500,000 from the fund is played out between the intermediate stages winners. The main stages of the competition IBM Watson AI XPRIZE (clickable)

    DARPA - Artificial Intelligence Exploration (AIE)

    DARPA announced the Artificial Intelligence Exploration (AIE) competition in July 2018 with the goal of creating a “third wave” of artificial intelligence. The ministry hopes that investment in science and technology development will bring positive results for the security of the state in the future.


    Researchers began working on the development of artificial intelligence back in the 70s. The first expert systems and servers for the execution of business rules DARPA calls the “first wave” of AI technologies.

    Later, neural networks learned how to perform classification tasks and make predictions - actively developing machine learning became the “second wave” in the field of artificial intelligence.

    Today DARPA seeks to create a “third wave” in the field of technology, the main characteristic of which is a symbiosis of robots and humans. Machines will no longer be programmed tools to perform their tasks - they will become colleagues and partners.

    AIE: goals and conditions The

    main task of the participants of Artificial Intelligence Exploration (AIE) is to develop artificial intelligence that can independently make decisions and adapt to different situations.

    Those who wish to participate will be periodically offered projects in various areas where it is necessary to develop AI.

    On September 28, the first draft was proposed.on AI development - Artificial Intelligence Research Associate (AIRA). You can present your solutions in two directions. One of them is AI algorithms that can discover new scientific laws and make equations for complex physical phenomena. The second direction is related to working with data and includes several tasks. Participants will try to determine which data is not appropriate to build models. You also need to learn how to determine whether the predictions of neural networks are correct, and evaluate the importance of experiments.

    To participate in the competition, you need to talk about your concept, describe the approach to work and present the purpose of the project. It is very important to provide a list of necessary materials and equipment and specify the desired budget.

    The organizers noted that projects that will help improve the work of the state will be sponsored. The maximum budget for each team is $ 1 million. DARPA will consider the proposals within three months - the selected projects will provide capital for research.

    Peter Heinem

    “We will select participants and enter into contracts quickly in order to see the result of innovative ideas as soon as possible,” says DARPA Deputy Director Peter Heinem.

    Within 18 months after the conclusion of a contract with a team of researchers, DARPA will be able to assess how far the project can be implemented and whether it will really contribute to the security of the state.

    AI Next Campaign

    The competition is only part of the AI ​​Next campaign, which was announced later - on the D60 Symposium in honor of the sixtieth anniversary of DARPA in September 2018. DARPA Director Stephen Walker said that the campaign aims to achieve three main goals, and creating a “third wave” of artificial intelligence is just one of the tasks.

    The second mission of the project is to learn how to use AI to enhance the state security system.

    Last but not least, the goal is to ensure the high-quality work of military structures with the help of artificial intelligence.

    The total amount allocated for investment in the development of AI in the framework of the entire five-year project is $ 2 billion.

    According to the official DARPA website, the key aspects of the study are:

    • new AI capabilities for automating US Department of Defense business processes;
    • improving the reliability of AI;
    • improved cybersecurity with ML;
    • reducing the amount of required data and electricity for training networks;
    • ML-algorithms of the new generation.

    With the help of the competition, DARPA plans to create a new generation of algorithms that will allow machines to independently argue and make decisions.

    Do contests decide anything at all?

    Against the background of the recent pitifully ended Lunar XPRIZE, doubts arise whether such contests help solve real problems? In addition, the created developments rarely find use immediately after their presentation.

    It seems that such contests have the greatest impact on the world indirectly - through the created information noise. The tasks assigned represent promising areas of work for researchers around the world. For example, when the competition for creating DARPA Robotics Challenge rescue robots was about to end, Honda announced the creation of its own rescue robot (and it was eventually created, but still remains very slow). Or a lot of environmental contests XPRIZE, which once again call to work on the problem (among the current ones - NRG COSIA Carbon XPRIZE).

    Even with the successful completion of the competition by 2020, the AI ​​created is unlikely to eliminate global problems. But maybe in another couple of years we will see some progress.

    PS In November, our two conferences on the use of artificial intelligence will be held.

    AI Conference Kyiv will be dedicated to automating business processes using artificial intelligence, IoT and chat bots. Representatives of Skylum Software, DataArt, Kyivstar, Ukrgasbank and others will speak at the conference.

    Moscow AI Conferencefocuses on the development, application and implementation of machine learning in workflows. Among the speakers are representatives of NVIDIA, Microsoft, CleverDATA, Yandex, and others.

    Also popular now: