29 June 2020

The US SOF are increasingly focused on AI

Daniel Ilie

Last year, I was writing that the US Special Operations Command – USSOCOM is working on its own AI (Artificial Intelligence) strategy, one to reveal their vision on how their SOF should adapt to the new challenges provoked by the fourth industrial revolution, which will change the nature of risks, vulnerabilities and threats against security.

Image source: US DoD

The Americans wanted to identify the constraints related to AI technologies’ expansion and machine learning (ML), to identify the purpose-driven application of these technologies, starting with the structures dislocated in the theatres of operations or the interested areas, up to the strategic level commands, meanwhile they were also allotting the necessary resources and were about to start the transformation of the human resources in a unit of digital experts. The human resource was about to identify the human-machine teaming opportunities, able to act on the entire spectrum of fight actions and features. Their purpose was for the US SOF to get, in 10-20 years, to fully use the AI and ML technologies as something common.

The foundation of the first US AI and ML strategy would be the so-called “365 SOF Artificial Intelligence Strategy, elaborated around three lines of effort, as follows: 1) preparing the labour force in AI field; 2) implementing AI applications; 3) the inter-institutional extension of AI implementation.

The identified interest areas were: perception and action, planning and execution of the manoeuvre, resilience of communication and cyber-defence, recruitment, instruction and management of talents, predictive maintenance, logistic planning and predictions and the management of contracts and acquisitions.

They were anticipating collective results such as: data and services based on cloud computing, the omnipresent use of “Agile” management practices in developing the classified and unclassified software products’ development, the advance of digital technologies acquisitions for the dislocated troops; attracting talents and training the labour force; the transition towards a more tenable digital future.

And, almost a year later, the USSOCOM commander, General Richard Clarke, was saying, during the first Virtual Conference of the SOF Industry, which took place last month, that “We’re going to have to have artificial intelligence and machine learning tools, specifically for information ops that hit a very broad portfolio, because we’re going to have to understand how the adversary is thinking, how the population is thinking, and work in these spaces”. He mentioned that, although traditional kinetic missions of SOF will not change too soon, other types of missions for these military capabilities, such as intelligence gathering, training and advising, and influence operations — need to be reprioritized.

It seems that USSOCOM started to use AI within the informational operations executed in the theatre of operations from Afghanistan, but they are also considering using it to develop applications in fields like the intelligence gathering and aggregation, surveillance and reconnaissance, execution and accuracy of the strike, health, and physical effort of SOF operators.

They obviously want to improve the informational cycle speed, allowing the analyst to focus less on normal tasks, like the gathering, reading of reports and making the logic connections between different data and information and use the time to understand the information. We already know that SOF needs the so called “actionable intelligence”. This is the critical intelligence which has an acceptable certainty level with limited availability, which, provided early and in proper time, is decisively contributing to special operations’ success.

This is also available for moments when, for example, according to USSOCOM, the SOF is in an “influence war” with enemies from a complex series of actors, like non-state group of actors, but also some which are developing informational operations funded by state actors. The Americans are concerned because their strategic enemies, especially those equally trained, are catching up or even surpassing them in terms of technologies and capabilities.

I do not know where are they with the elaboration of the first US SOF AI strategy, but they clearly know that without some concrete implementation measures it will not meet the expected results.

At the beginning of the month, it was created a new acquisition programs within the Program Executive Office: SOF Digital Applications/ PEO-SDA, part of the Special Operations Forces Acquisition, Technology and Logistics/ SOF AT&L, and USSOCOM body, which is responsible, since 1991, with all the research, development, acquisitions, technologies and logistics activities planned and executed for USSOCOM. According to officials’ statements, this acquisition programs will develop, provide and implement software capabilities and necessary data for SOF operators to win in the competition with the aforementioned enemies. This was possible after reorganizing the centre, both as a consequence to the gaps related to artificial intelligence development and the machine learning in favour of SOF, and a direction change of the command towards a software approach, eased by hardware equipment.

The new office wants to be a “unique counter” for digital applications and intensive software (a software that influences the design, setup, implementation and development of the system as a whole), where both SOF and the industry can share the capabilities they have. This will reach the initial operational capacity somewhere between 60 and 90 days from the elaboration, and the human resources will mostly come from the labour forces, not active military.

The program will include: intelligence systems for the SOF (following the structure of computer systems, able to process and disseminate information/ Distributed Common Ground System – DCGS SOF); systems to execute the centralized and decentralized command and take the common operation picture/ Mission Command/Current Operation Picture); management of the battlefield platform systems, mostly composed of a smart phone connected to a radio station/ Tactical Assault Kit –Core); systems for Special Operations Mission Planning and Execution; systems for SOF Planning, Rehearsal and Execution Preparation; digital ecosystems for SOF.

As for the use of AI and ML, the chief of the new acquisition program was saying that one of the challenges those who will enter the structure will meet will be to provide the interoperability of the autonomous sensors like unmanned vehicles or capabilities of gathering intelligence, based on space and cyberspace, but also integrating the intelligence provided by these and then distributing them to the final operator, who executed specific missions in the tactical field.

Also, attracting and keeping special talent and the necessary experience to solve such technical issues, on a competitive labour market, will be also a challenge. The USSOCOM hopes the industry representatives will join the process and help the implementation of AI and ML technologies for the US SOF.

The USSOCOM commander said, during the May conference, that “we need programmers” and that “We talked internally that maybe the most important person is not the operator that breaks the door, by the cyber-operator that the team will have to receive in the field, so that he can use its IT tools in the fight”.

In other words, do we need SOF tactical operators or programmers? This is the question when it comes to the competition between the more or less state actors in the so-called “war of influence”.

When it comes to such a dilemma,  I would says that getting the competition advantage in the future military confrontations will surely involve also using intelligent and unmanned armament systems, able to complete, eventually, the effort the human fighter put and to improve their performance and effectiveness in the battlefield.

However, SOF operators still need to have full control and initiative in the tactical field. The need for the human factor to have control over AI is still mandatory because those intelligent systems will have to reach intelligence levels long before being used, with strong confidence in the modern battlefield.

The ROU SOF will have to already think on how they will adapt to the new challenges, even if it is just for interoperability with the other NATO partners and to keep up with the technological developments. A national strategy in the national intelligence field would be adequate to that end. Unfortunately, its elaboration by Romania’s previous government, in 2019, did to not come to an end.

Translated by Andreea Soare