Artificial Intelligence and Dual Use in the Beijing programs

(To Antonio Vecchio)
24/09/18

We have already talked about the Chinese commitment in the field of Artificial Intelligence (IA) both at the experimental and the applicative level (v.articolo).

A phenomenon favored by the rigidly centralized governance of Beijing, which allows us to make long and very long-term plans, as demonstrated by Made in China 2025, the program that traces the path of the technological revolution now at the gates.

The Chinese military researchers are sure that the IA introduces a "singularity" in all the fields in which it will be applied, and push, supported by the current political leadership, to make China the global reference nation within the 2030.

Among all, the military one represents the area to which particular attention is paid, especially for the implications that derive from it on the role of future global player desired by Xi Jinping, according to whom "technology and science will strengthen military power ".

Before seeing what the state of the art of the military dimension is, and how Chinese military thought plans to use AI in future operations, it is good to understand what the Chinese mean when they talk about artificial intelligence.

The People's Liberation Army (PLA) dictionary describes AI as: "a weapon that pursues, identifies, destroys an enemy objective, consisting of data collection and management systems, knowledge based systems, assistance systems and decision-making support, and mission execution systems".

A broad and all-encompassing definition, which originates from the belief that the next operating environment will cease to be "informatized" as the current, that is characterized by a massive use of Information Technology Network, to turn into something totally different, "intelligented", in which the AI ​​will play such a preponderant role as to replace even human action.

The vast Chinese literature on the subject imagines future battlefields in which systems, and systems systems, "autonomous" will be employed1 and not autonomous, able to operate on multiple domains, with a wide spectrum of interventions in different sectors, not least that of deductive skill and the consequent decision-making ability.

The introduction of artificial intelligence in the military field is seen as a revolution, like the one introduced in the nineteenth century with steam and electricity, which entails - remaining in the military camp - the use of "intelligent systems directly on the hypothetical contact line, with the task of processing huge volumes of information and situation data, and so decide autonomously, and in real time, the lines of action to be taken"2.

One speaks therefore of a continuous cycle process, in which man - the "singularity" referred to above - will no longer be called to play, at least at the tactical level, a direct role (in the loop) , but only to ensure supervision over the whole process (on the loop).

The latter element, which makes a big difference in approach compared to the US counterpart, which never puts man out of the loop for value reasons (respect for human life) and for the profound implications of a legal nature (rules of engagement) .

At the moment the PLA studies in the field of AI concern:

  • autonomous and intelligent unmanned systems;

  • merging and processing of data;

  • processing of information;

  • defensive and offensive capacity in the "info-warfare" environment;

  • simulation, wargaming and combat training;

  • decision-making support.

In the continuous search for new fields of employment for the AI, the competition with the USA has a weight not secondary, for fear of losing the race in progress for the future "dominance" and not repeat what happened with the mechanization and the subsequent computerization of the PLA, which saw China settle largely behind the US.

This is probably the reason that pushed the Chinese from the study of the Third Offset Strategy, the main document with which the US Department of Defense outlined in the 2014 the defense capabilities, to be acquired using a wide use of robots, miniaturized machines and drones, and the exploitation of "big data" through the development of appropriate advanced calculation capabilities.

To get an idea of ​​the effort undertaken, just think that in the 2012-2017 period, Chinese investors spent 19 billion dollars, in 64 different development projects related only to the fields of robotics and virtual reality.

In this policy of technological innovation, unlike what happens in the USA, where the private sector is reluctant to cooperate with the defense sector for possible limitations to the commercial exploitation of products, China acts as a "country system", with a continuous exchange between industry - the three sisters BAIDU, ALIBABA, TENCENT (BAT) played a great role, and the military sector, under the direction of the powerful Central Military Commission led by Xi himself.

BAIDU, whose head of research comes from Google, is a global leader in the field of facial recognition (with the Apollo program) and unmanned machines (Duer Os program).

SENSE TIME, another big brand in the industry, is in the field of facial recognition, while IFLYTEK is a leader in that of speech recognition.

To this collaboration, all under the banner of dual use, the academic world is no stranger.

The university research is in fact moving in the field of Deep Learning, which consists in the research and application of algorithm classes for machine learning.

Also that of the Swarm Intelligence3, represents a field of research of great perspectives, having passed from the combined and synchronous use of 67 UAV in the 2016 to that of 1000's well 2017 machines. With huge relapses in the civil and industrial field, but also with the result of "tenfold" possible, future applications in the military one of long-range reconnaissance, jamming and "strike" (a swarm of UAVs could hit an aircraft carrier with devastating effects).

  

1A system is called "automated" when it acts primarily in a deterministic manner, always reacting the same way when subjected to the same inputs. An "autonomous" system, on the other hand, reasons on a probabilistic basis: received a series of inputs, elaborates the best answers. Unlike what happens with automated systems, an autonomous system, with the same input, can produce different answers.

To know more:

https://www.chathamhouse.org/sites/default/files/publications/research/2...

3Wikipedia: "swarm intelligence" (a swarm intelligence) is a term coined for the first time in 1988 da Gerardo Beni, Susan Hackwood and Jing Wang following a system-inspired project robotic. It takes into consideration the study of the systems self-organized, in which a complex action derives from a collective intelligence, as happens in nature in the case of colonie of insects or stormi of birds, or shoals of fish, or herds of mammals. According to the definition of Beni and Watt, swarm intelligence can be defined as: "propertiesà a system in which the collective behavior of (non-sophisticated) agents interacting locally with the environment produces the emergence of global functional patterns in the system". (Useful for "displaying" the concept: https://www.youtube.com/watch?v=axxXz2BM0yw)

(photo: The Verge / MoD People's Republic of China / TechNode / web)