Robotics & Geopolitics: Government Deals Could Affect AI Competition

Credit: iStock

July 26, 2018      

Countries around the world continue to work with other countries and companies as a way to cooperate on technology development, but also to protect their innovations. Government deals could influence the way robotics and artificial intelligence companies do business globally. This week, we look at a few agreements between nations, companies looking to protect their AI innovations from being stolen by others, and whether the U.S. can become a leader in industrial robotics.

Robotics Business Review has partnered with Abishur Prakash at Center for Innovating the Future to provide its readers with cutting-edge insights into recent developments in international robotics, AI, and unmanned systems. Are you ready to be updated?

Country agreements may influence robotics, AI landscape

Robotics development: Last week, organizations from Canada and Germany signed a Memorandum of Understanding (MoU) to cooperate in the fields of machine learning and quantum computing. The MoU calls for the creation of four groups that will focus on specific areas, such as machine learning and large-scale computing. These groups will include Canadian and German organizations that will cooperate on research. Part of the goal is to facilitate “international collaboration.”

Geopolitical significance: For robotics companies, it may be hard to see the importance of such deals. After all, how does a MoU between a small group of companies impact the wider robotics industry? Increasingly, governments and businesses are signing deals to cooperate in fields such as robotics. As this happens, it could change where the next competition comes from.

For example, earlier this month, during a visit by Chinese president Xi Jinping to the United Arab Emirates (UAE), the UAE and China called for enhancing ties in several fields, including in AI through technology transfers and data exchanges. At a time when the UAE plans to save $2.9 billion every year through AI (and blockchain), and when AI is projected to add $182 billion to the UAE’s economy by 2035, will Chinese AI firms benefit more than anyone else? If so, it could be due to country agreements that China and the UAE are signing.

Germany Canada government deals Memo of Understanding

Canadian and German research and private sector organizations come together to sign new agreement to establish Canada-Germany quantum computing and machine learning networks. Credit: Triumf

From the Middle East, look to Asia. In June 2018, Didi Chuxing, the Chinese equivalent to Uber (it actually bought Uber’s business in China), announced a new business in Japan, partnering with SoftBank. Part of this will involve using Didi’s AI to bring new ride-hailing services to Japan. However, this might not sit well with Japanese companies that are developing AI for taxis. In February 2018, Sony announced it would be testing AI taxi hailing in Tokyo this summer. In March 2018, Toyota announced it had developed AI that could distribute taxis more efficiently with 94% accuracy of where ride requests may be. In February 2017, NTT Docomo unveiled AI that can predict where ride requests will be. For one taxi driver, business increased 20%. As Sony, Toyota and NTT Docomo move to supply AI to taxis, the Japanese taxi market may face a new competitor: Chinese ride hailing services like Didi.

Geopolitics is underpinning these deals. To take Chinese AI into the Middle East, Beijing is working with Dubai and Abu Dhabi. Japanese businesses that want to overcome the bumpy China-Japanese relationship are identifying new sectors, such as intelligent transportation. These actions point to a new way for robotics firms to succeed. In order to succeed, companies and governments should build agreements that provide them with better access and more control.

Protecting AI from being copied or stolen

Robotics development: IBM is working on patent-pending technology that will let researchers protect their AI models from being copied. It works by “watermarking” the models, similar to the way many videos and photos are watermarked today. IBM showed a test of this technology at a conference in South Korea where it embedded a watermark in a deep neural network (DNN).

IBM Watermarking Government Deals article

Inventors and co-authors Jialong Zhang and Marc Ph. Stoecklin developed AI watermarking. Credit: IBM

Geopolitical significance: As international AI competition intensifies, governments and companies will seek new ways to protect their AI innovations. The technology IBM is working on may help, but it won’t solve everything. There is a high risk that companies working on expensive AI models could have their AI stolen by other countries.

For example, in June 2018, a U.S. startup working with the Pentagon on “Project Maven” was hacked, allegedly by people in Russia. According to a report, all the startup’s code may have been stolen. Separately, in September 2016, researchers in the U.S. and Switzerland showed that with access to just the Application Programming Interface (API), it was possible to reverse engineer algorithms with 99% accuracy. One of the researchers called it a “high school exercise.”

IBM AI Watermarking example government deals article

Model accuracy over training procedure. Credit: IBM

Equally important, there is complete ambiguity as to the rights and protections for content created by AI, such as music, videos, or photos. One of the few countries tackling this problem is Japan, which announced in May 2016 it would be developing a legal framework to protect “AI works.” Without a proper legal framework, robotics firms could face a future where their innovations are stolen. This may cause countries to intervene to protect their companies. This in turn may cause countries to clash.

As companies think about how to protect their innovations, another equally important question emerges: what happens after their AI is stolen? Perhaps robotics businesses will register patents in foreign markets, like China, to protect their AI. In January 2016, China’s Industrial Robot Patent Alliance launched the country’s first patent pool in Guangdong province. But what happens if their AI is stolen in Europe or South America? Even if local courts side with a company whose innovations have been stolen, it may be too late. The innovations would have been analyzed, shared and distributed long before any legal action.

Robotics companies need to think about their AI the same way governments think about their military hardware. In April 2001, a U.S. surveillance plane was grounded in China. By the time the U.S. received the plane back, it was stripped of everything. The U.S. knew the Chinese analyzed every corner of the plane to gain secrets. In the same way, robotics firms must think about their innovations like the U.S. surveillance plane, with the exception that companies should expect their innovations to be stolen.

The question in front of robotics businesses is once their innovation is stolen, stripped, distributed, shared, reverse engineered, manipulated, and resold commercially, how do they still maintain their edge? Answer that question and not only will a robotics company guarantee their future, but they will be guaranteeing their country’s future in the world of robotics.

U.S. aims to reduce dependency on foreign robots

Robotics development: By July 2019, Boston Dynamics wants its robot dog, SpotMini, to be available commercially. The company hopes to produce 1,000 units every year and to establish an ecosystem around the dog where developers can create new applications for it. According to Boston Dynamics, SpotMini is already being tested in construction, security, delivery and home assistance.

Geopolitical significance: For the U.S., its leadership has been in AI. When it comes to physical robots, besides things like robot vacuum cleaners, the U.S. doesn’t have the same leadership that Japan, Germany, Sweden and China have. This has made the U.S. dependent on automation technology from other countries.

For example, in June 2018, Kroger purchased a 6% stake in British online grocer Ocado. Why? Because Ocado has developed advanced logistics robots. Kroger wants these Ocado robots and has hired Ocado to build and manage 20 distribution centers. The deal between Kroger and Ocado points to something else: a U.S. supermarket leader is depending on British robots to compete. SpotMini may be the beginning of a new kind of U.S. leadership in the robotics world and the beginning of the world becoming dependent on U.S. robotics instead of the other way around.

The main way this will happen is if the U.S. government takes a leadership role. Could they look to Europe? In April 2018, the European Union (EU) unveiled a plan to create hundreds of “Digital Innovation Hubs” across the region. All the hubs will be networked together. The goal is to create European leadership in AI and not depend on China and the U.S. In the same light, the U.S. government could use SpotMini as a catalyst to help local robotics firms come out with new innovations so the U.S. can reduce its own dependency on foreign automation technologies.

For example, the next time President Trump meets with Japanese Prime Minister Shinzo Abe, could the U.S. include robotics as part of any deals that are signed, such as bringing SpotMini to Japanese construction sites? This wouldn’t be the first time U.S. robots were used in Japanese construction. In December 2017, Japanese construction equipment maker Komatsu partnered with NVIDIA to use AI to make construction zones safer. However, until the government unveils a robotics strategy, it will be up to private companies to lead. Is SpotMini the “push” the U.S. robotics industry needs?