Paper #35. What Might Military Operations Learn From Google? In Africa

Military operations must be prepared to conduct  so-called “great power competition” as well as big and small wars just as complex. In all cases, we need to implement superior strategy to defeat clever competitors. 

Especially with respect to competition in the cyber and information domain, it’s superior strategy that produces superior outcomes. The advent of Artificial Intelligence/Machine Learning (AI/ML) and quantum computing reinforces the importance of out-performing threats, particularly as AI learns to write its own code. 

In the recent book by Marco Iansanti and Karim R. Lakhani, Competing in the Age of AI: Strategy and Leadership When Algorithms Run the World, Google’s strategy to “put AI at the core” has implications for the warfighters in command and control of combined joint all-domain operations. Combined Joint All-Domain Command and Control (CJADC2) is focused on increasing the speed of decision-making for lethal effects. This is necessary, but insufficient to strategic success.

We need a language of strategy that wins all-effects wars, not just all-domain battles. To win all-effects wars, we need to deliver superior information for lethal effects when warranted, but more importantly, we need to shape non-lethal effects that last. Situational awareness of the latter is more difficult to realize. The outcomes include effective and fair governance, economic growth and prosperity, and social liberties and tolerance—all of which are diversely contested. As Iansanti and Lakhani explain, Google learns how to increase the value it delivers to customers via Artificial Intelligence (AI) that’s conversational, ambient, and contextual. I’ll unpack what these terms mean shortly. 

What happens if we apply this big tech business approach to a persistent security problem inflamed by lethality, past and present?

Consider Africa. 

Africa 

Africa has more multilateral peace operations throughout its territory than any other region in the world. Organizations include the African Union, European Union, Economic Community of West African States, Group of Five for the Sahel, Lake Chad Basin Commission, and United Nations. A variety of operations have emerged to address specific crises and conditions often regarded as intractable.

The continent faces a formidable array of economic, political and security challenges. African states seek private sector investment, expanded trade, debt relief and development aid—without losing sovereignty. Democratic reforms have displaced most overt dictatorships. However, ethnic-based patronage undermines growth and invites radicalization. There is widespread poverty, economic despair, political disenfranchisement, human rights abuse and a dearth of public services. Violent extremist organizations thrive in these conditions. 

Creating value-added engagement with the population is the best way to improve security conditions. This requires the kind of society-wide impact that Google has orchestrated by operating in the information environment. For military operations, this means putting Information Operations (IO) effects first. 

Putting IO first does not mean reducing lethal capabilities, as in air strikes against al-Shabaab bomb-makers who kill civilians, government officials and security forces. What it does mean is that IO needs to lead operations to win information advantage that defeats extremists’ control of the population. Nor is this an argument to increase US force deployments. Indeed, as of this writing the Trump administration announced the pullout of all US forces from Africa. Instead, this article assumes that the Biden administration will retain a footprint of US special operations forces (SOF) in Africa.

SOF are working with a variety of partners in separate and multilateral operations. Due to resource constraints, US Africa Command’s objectives have been downgraded from “degrade” to “contain” violent extremist organizations. So, let’s consider the effectiveness of the following two-fold approach. 

First, an “AI-first” foundation that creates conversational, ambient and contextual interactions. Second, operations that create and capture value.

Conversational, Ambient and Contextual AI

AI predicts what customers want or need. This is more than learning operator user preferences in CJADC2. First we need local data, secure software and devices to connect populations in meaningful ways. This challenge entails creating three key value-added processes.

• Conversational: provide the targeted population words and pictures that they want to share. This process informatizes operations. The task is to place data into local context, thereby creating relevant information for operations. Here are some questions to ask to guide collecting local data. What are the thick demographics? Namely, what are the lifestyles, routines, fears, interests, informal power relationships, and receptivity to types of change? What semiotics do locals use, and for what meanings?  
  
  The answers to these questions are unstructured data that can be ingested into a machine learning platform such as Savant X to visually display and update demographic interrelationships. This awareness can inform developers and operators about how to maintain conversations of local interest. Those conversations produce information, information that operations can be designed to influence for advantage.
• Ambient: all connected devices need to be compatible with the information of the conversation above. This process structures distribution. Mobile phones, personal digital assistants, tablets, or even watches with an operating system that supports apps can perform this function. However, only 18 percent of Africans have access to the internet. Nearly half of the world’s digitally unconnected population live in sub-Saharan Africa. In such remote sites, there is no substitute for human presence. Yet the correlation between sustainable development goals and mobile connectivity is clear:

• Contextual: embed desired changes that are meaningful to the population of interest. This process determines desired conditions. It’s making decisions about the desired set of conditions (end-state) that flows from the ambient conversation processes above. These conditions are a blend of the status quo and desired changes.
  
  Unconnected villages must be included in conversational AI threads, but via real people rather than devices. The only ambient ”devices” may be the human inhabitants, until an ink-spot of digital connectivity happens.

With these processes in place to informatize operations, structure their distribution, and purpose them to set conditions, we have to incentivize and realize value.

Create and Capture Value; Operate

The next challenge is to create and capture value, and put that into operation with activities.

Businesses create value in terms of product or service loyalty when they provide what customers want in terms of price and quality. What is the price that Africans pay to get what they want, or to avoid what they don’t want? Lives and lifestyle disruption are the price urbanites and villagers pay to be secure. Security partners have to provide reliable security at an acceptable price. Ideally that’s zero lives lost and lifestyle preserved. The enemy competes with weapons of intimidation and coercion, increasing the price (accommodate or fight the enemy) that Africans pay for protection. Security partners can dominate this space by providing a higher quality service.

Conversational, ambient and contextual information increases awareness of the value of a security partnership. With leadership, that deepens the mutual commitment of security partners. 

Businesses capture value by monetizing what customers want for a profit or to gain market share. An isolated African village will have to pay more for security protection. So in return, the village must receive more security than the enemy can provide. That higher price is engagement with security partners in terms of providing intelligence, participating in a neighborhood watch or militia supported by competent and reliable security partners. 

What does this add up to so far? Winning an information contest requires a competitive commitment to quality and engagement that is valued by the population. Conversational, ambient and contextual information technology raises expectations among Africans about a better life. This increases the likelihood that more people are willing to pay the price for meaningful security, with hope for the future. 

Businesses put value creation and value capture into operation with a variety of strategies to compete against alternatives.  Among connected stakeholders and partners, AI-centric information technology confers an advantage of proactive decision making. Let’s see how.

Application to Somalia

To provide some challenging details, consider Somalia. My recommended security strategy priorities for Somalia, explained in ICSL Paper #33, are fourfold: 

• Dissuade Somali youth from being recruited by extremists
• Persuade the Somali government to develop a narrative for popular support
• Coerce extremists from weaponizing information and acquiring lethal weaponry
• Induce Somali government initiative in planning, reconciliation and reform

These priorities are ambitious and difficult to implement. Here is a simplified, partial version of the eight strategies that operationalize the four priorities. The strategies are combined strategies in that they complement one another for synergistic effects. The activities that comprise the operations which implement the strategies interact among one another. This is where AI can help.

The reason for the level of detail below (simplified from the full version of eight strategies) is to suggest where a conversational, ambient and contextual AI can help humans make holistic strategic decisions that out-compete threats. Specifically, consider how AI could diagnose options to adjust these interrelated operations. A few ideas to do so are bolded in the tasks underneath each strategy:

Combined Strategies and AI Tasks

• Dissuade youth from being recruited by extremists by protecting moderate Islamic leaders
  • AI: visually map youth preferences and influencers on recruitment issues, indicating their relative weight and direction
• Persuade Somali army and police to develop a narrative that undermines extremism and assures citizens’ rights under a national judiciary
  • AI: identify vulnerable elements of extremist narratives and related stories, cross-referenced to malicious patterns and local attitudes
• Induce Somali army and police to expand professional networking and engage the African Union’s mission by hosting military planning via secure mobile platforms
  • AI: maintain vulnerability assessment, and predict physical and virtual attacks, with live red-team threat profiles of extremists
• Defend the populace from extremists’ extortion of taxes via tracking, and by providing reliable basic government services
  • AI: anticipate victim contact-tracing and government services expectations with automated analytics
• Induce dialogue and reconciliation among the Somali federal and state governments with joint command and control of national operations
  • AI: predict key relationships and related joint operations opportunities against threats
• Induce the Somali government to implement political reforms for debt reduction,  governance and anti-corruption
  • AI: illustrate benefits of reforms to individuals/groups in terms of immediate conditions and long-term alternative futures; include practical options to compensate influencers
• Defend against extremist recruitment in-person and on-line by denying movement and neutralizing anonymous influence
  • AI: predict enemy physical movement and psychological maneuver in geographic and virtual space
• Coerce extremists into defecting by denying seizures of weaponry and revealing unjust casualties and violations of Sharia
  • AI: generate information vulnerabilities of specific groups and leaders based on integrated human and environmental factor profiles

Conclusion

As illustrated by each AI task, each strategy can be enhanced with population-oriented information. AI can engage targeted audiences with accessible technology that influences behavior. Human engagement does not go away. Rather, it becomes AI-informed human engagement. Companies such as neoeyed can provide behavioral analysis, conversational AI, and red-team machine learning.

The challenge in using AI is similar to connecting CJADC2 platforms of sensors and shooters, but with a broader perspective of sensors and influencers. In complex environments, influence subsumes destruction. The author’s framework of cooperative and confrontational effects, for instance, includes persuade and dissuade, compel and deter, induce and secure, coerce and defend.

All of this activity is guided by humans in control. We control AI by asking the questions we want to ask. Human-controlled AI can deepen mutual commitment and raise expectations, which is risky if there is no follow-through. Security partners can create and capture that new value with proactive decision making. Again, what about the “unconnecteds”? 

Extremists in Somalia, primarily al-Shabaab, ban internet usage. Only 10% of Somalis, max, have internet access. Therefore the unconnected vast majority of the population is influenceable indirectly, not directly. As threats change, it’s the information environment that shapes behavior. Uncertainty is pervasive, and all the more exploited by adversaries in an IE that is not contested. To reduce threats, the population has to be protected from extremists, which requires physical presence.

Activities and operations must influence the enemy and protect the population. This requires will and capability. Security forces may have superior capability, but that needs to be smartly implemented to overcome superior will. Google’s success demonstrates that conversational, ambient and contextual intelligence—artificial or human—influences will.

Advantageous information effects, such as those in the eight combined strategies, must lead operations. Why? Because the whole point of these operations to dissuade, persuade, induce, defend, and coerce.

We should learn how military operations, in combination with other instruments of influence, can create lasting information advantage. We would learn the strategic purpose of fighting battles.