Giancarlo Elia Valori
Over the past two years, the development of Artificial Intelligence and the new techniques for using Big Data has become both faster and more widespread.
According to the old definition, by Artificial Intelligence we mean teaching a machine to think like a man, while Big Data is such a large mass of data in terms of quantity, speed and variety that it has to enable specific technologies and methods to extrapolate data from news already learned and extract new data and links from the news which seem unrelated to one another.
This ranges, for example, from the analytical forecast of buyers’ behaviours – by always using machine learning – to the inference of relations between single data and sequences of phenomena. Just to make an example, each buyer wants a specific reward.
Currently we also have the possibility of developing Generative Adversarial Networks (GANs), which create objects not existing in reality, but similar to reality, as well as faces that have never been seen before but are quite probable, and objects that do not exist but seem to work well.
Not to mention the self-correcting systems based on concepts that are adapted by the machine itself, as well as programs that self-create themselves, starting from a small nucleus.
In the United States, the total investment in AI companies is already worth 2.3 billion US dollars.
According to the analysts of this specific market, however, there are some trends which will emerge shortly and will make the difference among the various global competitors.
Reinforcement learning, for example, is a technique enabling the software used to maximize a cumulative reward. An information reward or even a reward in terms of speed in data search.
A sort of Pavlovian algorithm favouring the most suitable AI network and, above all, more capable of creating new algorithms during its “evolutionary” activity.
This AI device is needed for robots, but – as can be easily imagined – also for university training or for health, especially for arranging therapies for chronic diseases or even for analyzing and forecasting the share flows on the markets.
There is also Artificial Intelligence for quantum computing, i.e. a technology used by the computers operating with quantum physics.
Besides processing information in the “classic” way, quantum computers use two specific characteristics of the quantum system, i.e. overlapping – where two or more quantum states can be added together – and entanglement that implies, in a counter-intuitive way, the presence of many remote correlations among all the physical quantum states examined.
Hence an availability of data and calculation speeds, enabling to carry out previously unimaginable operations: the analysis of continental climate change; the world economic cycles of raw materials; the number and physical constants of galaxies in space.
In the future, there will also be convergence between AI and the Internet of Things, which will make both the construction of vehicles and their driving autonomous.
Another short-term integration will be between blockchain technology and Artificial Intelligence.
We have often spoken about blockchain, but in this case it is above all the integration between the blockchain “closed” network and a selective data collection or, otherwise, a patented and still secret technology.
Another promising AI sector is facial recognition, as well as the specialized programs’ ability to recognize manipulated data.
We can easily imagine to what extent this algorithm is important in intelligence analysis.
We will also have complex neural networks available for “deep learning” but, above all, we will have the possibility of developing very complex and highly predictive socio-economic models.
Deep Learning is the AI automatic learning network using concept or sign hierarchies, where higher-level concepts are defined by lower-level concepts. Identifying the genetic sequences of some diseases, as well as identifying tumours with X-ray and arranging an automatic supermarket are all Deep Learning operations.
However, there will also be significant developments in privacy protection and in the development and processing of natural language, both for Deep Learning, which often uses personal data, and for the other AI techniques.
Hence how do major countries act, faced with this new extraordinary technological and productive opportunity?
China has entered the global AI “first level” as early as 2017, while it sells many weapons with Artificial Intelligence content in the Middle East (Saudi Arabia and the United Arab Emirates) and in the areas where it is not possible to trigger competition between China and the other countries having AI technologies.
For the Chinese People’s Liberation Army, war is currently shifting from the destruction of the “conventional” enemy to the operations for harming and eliminating the enemy, which are based on AI, but are extremely fast and aim at the enemy’s complete destruction.
For China, in the future war will be a “confrontation of algorithms” and not a clash of “forces”.
In addition, President Xi Jinping and his team believe that, as the role of AI is expanding in both the civilian and military systems, China must rely ever less on imported technologies and ever more on those developed within China.
Precisely in October 2018, President Xi chaired a special Politburo on AI.
Hence strategic, scientific and technological self-centredness, together with the achievement of world hegemony.
An important strategic element in the Chinese AI doctrine is the need – raised by some leaders as early as 2018 – to “avoid the AI global threat” and hence set some global checks at multilateral level, as happened for nuclear and chemical weapons.
A recent document drafted by the China Academy for Information and Communication Technology already speaks openly about international standards that can put AI under control.
Furthermore, the Chinese military decision-makers are already thinking about a future war “without fighters”, with weapons fully independent from man and even transported autonomously.
Currently China is already exporting most of its aerial drones to the Middle East, including the latest generation ones, which are almost all remote-controlled.
China, however, also shows strong interest in military robotics and, particularly, in automated military decision-making.
In China’s current doctrine, there is – first and foremost -intelligence supremacy, which is almost naturally followed by AI dominance.
In Xinjiang, for example, Artificial Intelligence is already used against local terrorists.
In this case, AI technologies are used to identify and track all terrorist activities, both through the sensor network and by means of facial recognition technologies and the recognition of other physical characteristics.
Moreover, the Chinese government has established two new research centres in the AI field, namely the Unmanned Systems Research Center and the Artificial Intelligence Research Center.
They are dedicated to the AI dual use research – both civilian and military research – but, despite other countries’ undeniable AI development, China wants to become the top country in the field of research, AI patents, venture capital invested in AI and number of companies dealing with Artificial Intelligence. Finally, it wants to become the largest pool of talent in the world.
With specific reference to the analysis of its own strategic weaknesses, currently China perceives it has AI limits in terms of best researchers; technical standards used; the quality of software platforms and the evolution of semiconductors – which is essential for developing advanced software systems.
Chinese leaders find other technological limits of their country’s AI project in the specific hardware for AI platforms and in the evolution of algorithms.
Currently the “best” AI experts are approximately 204,575 worldwide.
The United States currently has at least 28,536 of them and China, which is already ranking second, has over 18,232 of them.
China, however, is still ranking only eighth in the list of Top AI talent, with mere 997 AI scientists at the highest levels, compared to 5,518 in the United States.
With specific reference to the search for new AI technologies and markets, in its official documents China argues that we should always “abide by market mechanisms, but step up the marketing of AI technologies to create a comparative advantage. Finally, Chinese operators must always well understand the division of labour between the market and the government.” Marketing to create a comparative advantage is a very interesting concept to evaluate China’s Artificial Intelligence strategy.
For China the turning point will be the development and autonomous innovation in the semiconductor industry, which currently – as in the past – is at the core of information technologies, at first, and later of AI technologies.
In the future, the new AI technologies will be quickly marketed in China to support the financial effort for their implementation and, above all, to keep on operating with the old mass tools and instruments in the intelligence field.
In abstract terms, however, which are the factors of national power in the AI era? Firstly, a large amount of useful data must be available.
In fact, AI will greatly increase the power of the countries capable of identifying, acquiring and applying the data sets enabling to develop new effective AI architectures.
More data, more algorithms. More algorithms, more accuracy and complexity.
Furthermore, considering that the abilities for developing AI are still very rare in the research community, the Artificial Intelligence competition will be won by the country that will invest a great deal of resources in research but, above all, in the salaries and scientific equipment of AI researchers.
Nor should we forget the resources for calculation, which must already be very large.
Currently the most powerful computer in the world is already Chinese.
Then there must also be the political or economic incentive to adopt AI in business, in companies or in offices.
With specific reference to investment, a sound correlation is also needed between the private and the public sectors.
China is favoured from this viewpoint, considering its link between the military and scientific academies, while the United States shows some limits.
These limits are inherent in the Silicon Valley’s private operating logic and in the scarce relations between it and the military decision-makers.
Finally, however, there is also a smart – and not mythological – assessment of privacy regulations.
The countries that – sometimes obsessively – give priority to privacy over other regulations, are obviously slower in developing advanced AI technology.
Here again material technology is decisive in itself: the evolution in graphics cards, the chips with particular characteristics and the evolution of hardware enable to achieve not the abstract possibility of Artificial Intelligence, but its operational existence.
Without a specific level of technology already reached, AI is simply impossible.
Apart from the United States and China, Israel invests in AI for both commercial and military reasons: currently Israel has already collected a total amount of 7.5 billion dollars to invest in AI, with 950 small companies, 51% of which use machine learning technologies.
The Russian Federation has long been investing a great deal of resources in AI and robotics.
Last year Russia even doubled its AI investment while, according to its military leaders, “robots will be the real protagonists of the future war”, while Russian military staff is already tending to the “complete automation of military space “.
The well-known Kalashnikov company has already studied and marketed a series of autonomous weapons, managed by AI neural networks. In the near future, however, there will be Russian robotic nuclear submarines, in addition to the Armata T-14 tank – also incorporating AI technologies -which has already been used in Syria.
Here the legal matters to which China sometimes refer are linked, above all, to the Convention on Prohibition or Restrictions on the Use of Certain Conventional Weapons which may be deemed to be excessively injurious or to have indiscriminate Effects, a Geneva Convention of 1981 signed by 50 States.
While the aforementioned Convention applies to the new killer robots and lethal autonomous weapons systems (LAWS), Russia, the United States, China and Israel must obviously put new types of LAWS into action.
They are certainly not robots like those of the 1960s comics, but conventional weapons, at least apparently, which decide for themselves – without human command and control – who must live and who must die.
China, however, agrees with the new LAWS criteria, but the fact is that: a) LAWS weapons are decisive for the future battlefield; b) China has already decided and established the technologies suitable for the future LAWS; c) China has developed even more advanced weapons, in relation to the new growing powers.
The new arms race will always take place within the AI context.
With specific reference to Russia, however, additional considerations must be made.
In 2014, Russia’s political and financial system defined 9 high-tech sectors, in view of Russia producing these AI technologies by the end of 2035.
The projects are AutoNet, AeroNet, EnergyNet, FinNet, FoodNet, HealthNet, MariNet, NeuroNet and SafeNet, which are basically all AI networks.
So far 1,400 AI projects have been carried out in Russia. According to the Russian government’s forecasts, the AI and machine learning market is expected to increase by at least three times within 2020 while, over the next five years, 80% of decisions in financial markets will be taken through AI, while 50% of the service sector will still be dominated by AI techniques, in both the field of e-commerce and of other types of trade.
There is also Singapore, a small but powerful hub for Artificial Intelligence.
Finally, there is South Korea, which uses AI for its financial and export markets but, above all, to control the Demilitarized Zone on the border with North Korea.
Hence, in principle, we could say that AI operates preferably in capital-intensive countries.
Certainly, with specific reference to the AI military issues, in the United States there is Google, which can be partly used by strategic networks, but China has full State control of the Internet, which allows a huge collection of data to later process some “useful” algorithms.
In principle, we have already seen China’s future AI strategy.
Conversely, so far the United States has not had a real national AI strategy, although currently – after some documents of the White House during Barack Obama’s Presidency, but above all after some decisions taken by President Trump – AI has been integrated into the Defence sector and is examined in relation to the possibility of creating a private market leading to the victory of the US Artificial intelligence over the Chinese or Russian ones.
Among the countries which are less interested in or capable of achieving AI hegemony, there is also India, which is interested in the Artificial Intelligence applications in the agriculture and administration sectors, while it deals with automated land vehicles and robotics in the Defence field.
In April 2018 the European Union developed a “Strategy for Artificial Intelligence”, with 20 billion euros of public and private investments until 2020 and additional 20 million in the following decade.
There will also be a group of EU experts, although we do not know yet who will appoint them. This group will deal with the ethical guidelines for the use of Artificial Intelligence.
The Italian public investment for the Internet of Things and Artificial Intelligence must be placed within this framework. In the public budget for the period 2019-2020, this investment is expected to be 15 million euros. There is no need for further comments.
Honorable de l’Académie des Sciences de l’Institut de France
Presidente di International World Group