20.07.2018 - Natural Sciences Sector

Towards a monopolization of research in artificial intelligence?

© Technion Institute of Technology, This tiny robot developed by the Technion Institute of Technology in Israel can theoretically be guided inside the body to perform a variety of medical tasks.

Since the release of the UNESCO Science Report in November 2015, the rapid development of artificial intelligence (AI) and the implications of these technologies for national innovation systems and the wider economy have led a growing number of countries to develop strategies for AI.

Research and development (R&D) in AI are being driven by multinational corporations headquartered in a handful of countries. This concentration of research is fuelling a fear of being ‘left behind’. It raises the spectre not only of a widening technological divide between developed and developing countries but also of growing dependency on a few commercial powerhorses.

When the Organisation for Economic Co-operation and Development (OECD) studied the patenting behaviour of the top 2 000 companies for R&D between 2012 and 2014, it found(1) that corporations headquartered in just seven economies accounted for 93% of AI patents registered with the top five patent offices(2): Japan (33%), the Republic of Korea (20%), USA (18%), Taiwan, China (9%), China (8%), Germany (3%) and France (2%).

The OECD study found patenting to be more dynamic in AI than in any other technology, with the number of AI inventions increasing by 6% per year on average, twice the average annual growth rate observed for patents overall.

In some technological fields, AI-related components are already becoming commonplace. For instance, up to 30% of patents filed in medical diagnostics between 2012 and 2015 included AI-related components, according to the OECD study.

The same economies dominate both AI patents and industrial robots. In 2016, five economies accounted for 74% of global sales in industrial robots: China (excluding Taiwan China, the sixth-biggest market), the Republic of Korea, Japan, USA and Germany.(3) The most automated countries in the world are the Republic of Korea, Singapore, Germany and Japan.

‘Digitalization will affect all countries, irrespective of whether they actively pursue it’, concluded UNCTAD in its World Investment Report 2017 on the digital economy.(4) ‘Developing countries, and especially least developed countries, may risk increasing dependency on a few global digital multinational corporations, or further marginalization from the global economy’.

The risks of concentrating power in an economic elite

In 2017, Cockburn et al.(5) argued that the potential commercial reward from AI research would give individual companies powerful incentives to acquire and control critical large datasets and application-specific algorithms’.

Yoshua Bengio, an expert in deep learning, expressed similar concerns in the July 2018 issue of the UNESCO Courier. ‘Economists predict that AI products will account for up to 15% of the total production of goods within a decade’, he said. AI will allow companies ‘to sell more, to get rich and to be able to pay the researchers they recruit even more than they do now. By increasing their customer base, AI companies ‘will increase the amount of data they have access to – and that data is a goldmine that makes the system even more powerful…’

‘Such a concentration of power can have a negative impact on both democracy and the economy’, he said. ‘It favours large companies and slows down the ability of small new companies to enter the market, even if they have better products to offer.’

Bengio made a plea for international coordination to counter the monopolization of research in AI. He also evoked the need for anti-monopoly laws. ‘History teaches us that they can be effective against the excessive power of some companies’, he remarked.

Cockburn et al. concluded from their own study that the adoption of policies encouraging transparency and the sharing of core datasets across both public and private actors could stimulate innovation-oriented competition and enhance research productivity.

A flurry of national strategies for artificial intelligence

A growing number of countries have adopted, or are working towards the adoption of, national AI strategies. Policy researcher Tim Dutton is regularly updating a list(6) that includes the following: Australia, Canada, China, Denmark, Finland, France, Germany, India, Japan, Kenya, Republic of Korea, Singapore, United Arab Emirates, the UK and USA. The European Commission is working with member states to develop a coordinated a strategy for AI by the end of 2018.

Dutton has observed that ‘no two strategies are alike, with each focusing on different aspects of AI policy: scientific research, talent development, skills and education, public and private sector adoption, ethics and inclusion, standards and regulations, and data and digital infrastructure’.

For instance, in July 2017, China released its Next Generation Artificial Intelligence Development Plan. The plan’s stated aim is for China to attain the same level in AI as the USA by 2020 and become the world leader in AI by 2030(7). Bolstered by generous public-sector funding, Chinese start-ups in AI numbered 383 by 2017. This is still well behind the world leader, the USA (1 393), but ahead of Israel (362), the UK (245) and Canada (131). The Republic of Korea ranked twelfth, reflecting its lesser entrepreneurial culture.

National strategies for AI are building on past efforts to digitize industry. As the UNESCO Science Report recalls, Germany’s high-tech strategy has emphasized cyberphysical systems since 2011, through the Industry 4.0 project, a reference to the fourth industrial revolution. China has set up an Internet of Things Centre and India a Cyberphysical Systems Innovation Hub. In 2013, the USA launched the Advanced Manufacturing Partnership and, a year later, Canada added advanced manufacturing to its federal priorities for science and technology.

With cognitive sciences being a vital component of AI, the European Union, Japan and the USA have all launched major projects since 2013 to study the human brain. In March 2018, the Chinese government announced its own initiative, the Chinese Institute for Brain Science, Beijing.

The current concentration in AI is part of a wider trend

The current concentration in AI is part of a wider trend. Over the past decade, there has been a dizzying number of mergers and acquisitions which have created powerful corporations with a bigger GDP than some national economies.

The figures speak for themselves. ‘In the USA, 12 249 companies concluded mergers and acquisitions in the twelve months to 30 June 2015’, tallies the UNESCO Science Report. Of these deals, 315 represented an investment of more than US$ 1 billion. ‘Notable among them was a flurry of acquisitions by technology giants Yahoo, Google and Facebook, each seeking to add new talent and products to its stable’.

Of the top 100 digital multinationals, 60 are based in the USA. According to the US National Science Foundation(8), between 2011 and 2016, the two US industries that received the largest amount of seed-stage investment were software as a service ($3.8 billion) and mobile ($3.5 billion). Artificial intelligence, consisting of a variety of technologies that include software, natural language processing and optical character recognition technology, received $0.8 billion and life sciences $0.5 billion.

State-owned enterprises are merging, too, to help them compete globally. The UNESCO Science Report relates how the Russian government decided to amalgamate more than 90 state-owned industrial enterprises and research centres into the single United Rocket and Space Corporation in 2013. Two years later, this corporation merged with the Federal Space Agency to form the Roscosmos State Corporation, which concentrates R&D, manufacturing and land infrastructure. In April 2018, Russia Today announced that Roscosmos was planning to merge its Glonass satellite tracking system with the Chinese system Beidou to create a single navigation system which would cover much of Eurasia, for use by members of the Shanghai Cooperation Organization.

Firms in emerging economies have embarked on a series of cross-border mergers and acquisitions, in order to reduce their dependence on foreign core technologies. Mergers and acquisitions can be a way to acquire skilled personnel, patents and a sophisticated product catalogue overnight.

Tata launched India’s first wave of cross-border acquisitions with the purchase of The Corus Group plc (today Tata Steel Europe Ltd) in 2007. This purchase gave the Indian company access to car-grade steel technology, a key advantage for an industry that is India’s second-biggest after pharmaceuticals.

The state-owned China National Chemical Corp (ChemChina) took over the Swiss agrochemical giant Syngenta in 2017, in the biggest foreign purchase yet by a Chinese firm.

China has become the world’s biggest market for industrial robots, with a 31% share. In 2016, the Chinese company Midea purchased Kuka, Germany’s biggest and most advanced manufacturer of robots. Midea had been purchasing robots from Kuka since 2011, during which time it had reduced the size of its workforce by half.(8)

Multinationals relocating research abroad

Multinational corporations are taking advantage of globalization to move their research activities overseas. As the UNESCO Science Report observes, they may be motivated by a combination of factors, such as attractive tax breaks, the availability of local talent, the opportunity to streamline the speed with which they get their products to market, or the chance to adapt existing products to a new market.

Some multinationals have gone so far as to change their corporate citizenship. Through a process known as tax inversion, US companies may legally acquire a firm in a country with a lower corporate tax rate, such as Ireland. Ireland’s 12.5% corporate tax rate has made it an attractive destination for tax inversion, a trend which may explain why corporations headquartered in Ireland rank 15th worldwide in the OECD study of AI patents.

In 2015, for instance, US pharmaceutical giant Actavis, which makes Botox, acquired the smaller Irish company of Allergan, adopting Allergan plc as its new global name and Dublin as its new home. Less than a year later, pharma giant Pfizer attempted to acquire Allergan plc in what would have been the largest tax inversion ever but the deal was shelved following the adoption of new rules by the US Treasury.

The UNESCO Science Report observes that ‘there are emerging concerns in Europe about the erosion of its science base through takeover bids from competitors’. One illustration of this is Pfizer’s aborted takeover bid of the Anglo-Swedish pharmaceutical company AstraZeneca in 2014, after it failed to convince the British government and AstraZeneca’s board of directors that the takeover would not affect research jobs in the UK. In August 2018, the German government vetoed the purchase of the domestic firm Leifeld by the Chinese Yantai Taihai Group, citing security reasons. The German company manufactures specialized machines and equipment for industry(9).

Most multinationals, however, content themselves with relocating their R&D abroad to take advantage of specific local skills or markets. In India, the information technology (IT) industry is dominated by multinationals headquartered mainly in the USA. These foreign research units take advantage of the skilled, yet cheap labour available in India in software engineering and applications. By 2013, 93% of the Indian utility patents in IT registered with the US Patent and Trademark Office (USPTO) were owned by foreign firms.

The rapid growth in foreign ownership of IT firms has meant that foreign patents in this field now dominate India’s high-tech industry, ahead of India’s leading home-grown industry, pharmaceuticals. Indian pharmaceutical companies accounted for one-quarter (28%) of the country’s industrial expenditure on R&D in 2010, almost three times the amount spent by Indian IT companies.

Unlike in IT, almost all of the pharmaceutical patents secured by Indian inventors with USPTO belong to domestic companies. Nevertheless, pharmaceuticals accounted for just 9% of Indian utility patents in 2012, compared to 61% for IT. Fifteen years earlier, the situation had been reversed, with 31% of Indian utility patents relating to pharmaceuticals and just 12% to IT.

The implications for domestic industry

The case of Israel offers an interesting perspective on the interaction between foreign multinationals and domestic industry. Israel is considered one of the biggest centres for venture capital outside the USA. Yet, even here, foreign patents dominate the domestic market. Since 2002, almost 80% of applications to the Israel Patent Office have been filed by foreigners, according to the UNESCO Science Report. A sizeable share of these foreign applicants are pharmaceutical companies, including F. Hoffmann-La Roche, Janssen, Novartis and Pfizer.

The foreign research centres implanted in Israel tend to be primarily owned by US firms, many of them digital giants such as Qualcomm or Google. Within a year of the founding, in 2014, of CyberSpark, an Israeli innovation hub specializing in cybersecurity, several Israeli start-ups there had been acquired by foreign multinational companies, including Intellinx, purchased by Bottomline Technologies, and Cyvera, purchased by Palo Alto Networks.

The UNESCO Science Report concluded from this that ‘the loss of intellectual property into the hands of multinationals occurs mainly through the recruitment of the best Israeli talent by the local R&D centres of multinational firms. Although the Israeli economy benefits from the activity of the multinationals’ subsidiaries through job creation and other means, the advantages are relatively small compared to the potential economic gains that might have been achieved, had this intellectual property been utilized to support and foster the expansion of mature Israeli companies of a considerable size.

Source: Susan Schneegans, with excerpts from the UNESCO Science Report: towards 2030 (2015).

___________________________

(1) OECD (2017) Science, Technology and Innovation Scoreboard 2017: the Digital Transformation.
(2) Patents registered with the top five patent families: European Patent Office, Japanese Patent Office, Korean Intellectual Property Office, State Intellectual Property Office of the People’s Republic of China and the US Patent and Trademark Office.
(3) According to the International Federation of Robotics’ 2017 report.
(4) UNCTAD (2017) World Investment Report 2017: Investment and the Digital Economy. United Nations Conference on Trade and Development, Geneva.
(5) Cockburn, I. M.; Henderson; R.; and Stern S. (2017) The Impact of Artificial Intelligence on Innovation. See: http://www.nber.org/chapters/c14006.pdf
(6) See https://medium.com/politics-ai/an-overview-of-national-ai-strategies-2a70ec6edfd
(7) Asgard and Berger (2018) Artificial Intelligence – A Strategy for European Startups. Recommendations for Policy-makers.
(8) NSF (2018) Science and Engineering Indicators. National Science Foundation, Washington D.C.
(9) Deutsche Welle (2018) Chinese takeover of Leifeld collapses ahead of expected German veto.




<- Back to: All news