If democracies are to compete with savvy authoritarians, we need to up our game in the artificial intelligence (AI) information environment (IE), where out-thought is outfought. Beyond a profession of arms with all-domain military strategy, we need a profession that integrates all effects. Such a profession of effects begins with strategy.
In an AI-influenced IE, all elements of strategy–ends, ways and means–need the agility to change will and capability. Our operating concepts should adapt to current and anticipated conditions. As a process, strategy applies to diverse circumstances where it competes for relevance. Industrial age concepts of “initiative,” for instance, operate differently in the contemporary IE with respect to tempo, mass, learning, decisionmaking, position, and freedom of maneuver (see Paper #20).
Understanding historical events in their contexts can develop judgment for strategy in different conditions. Historians write about strategy in terms of human agency, contingency, and context. The emphasis is on a situation’s distinctiveness. That deep description can provide insights for that context, but we also need effective strategy suitable to changing and uncertain contexts. So, based on our historical understanding of current and anticipated conditions, what do we assume about a situation until it’s disproven?
American strategic failures—notably Vietnam and Afghanistan—are results of unrealistic assumptions about strategy. Among the various missteps of mirror imaging one’s adversary or partner, the critical mistake is to equate peace with cooperation and war with confrontation. This tendency oversimplifies strategy as it fits our binary, idealistic narrative of liberal democracy. We assume that we’re either in peace–where we deter or in war–where we fight.
American leaders speak to this narrative by endorsing a “when-deterrence-fails” approach to warfare (see Paper #40). So we follow the law of armed conflict, but those are narrower than Geneva Law or Humanitarian Law. That gap leaves space for authoritarian competitors to wage all-effects all-domain warfare that we don’t recognize as warfare.
For many authoritarians, there is no such fantasy assumption about peace and war (see Paper #38). Russian and Chinese active measures combine territorial invasions and economic coercion with systematic disinformation and diplomatic persuasion. Our competitors see the world in a perpetual state of peace-war, where talk-fight and coop-frontation are normal.
Democracies need superior strategies with realistic assumptions while preserving our ideals. A profession of effects that’s accountable as an institution and competitive as a system would be broader than the profession of arms. Accountable professionals working in teams could provide elected and appointed officials entrusted with proper constituted authority more options for consideration. That includes competing and waging warfare in kind–diplomatic, informational, military, economic, and social. The core competency of such a profession should be in strategies that subsume the current consensus on dealing with threats: brute force and coercion.
American strategic thought since the Cold War is based on those two options. Brute force is for warfare, while coercion, defined as compellence and deterrence by denial and punishment, is for peacetime. Because coercion is not brute force, coercion theorists regard compellence and deterrence as cooperation. After all, coercion theory emerged out of a sincere desire to deter nuclear war, which has worked so far.
The problem is, today’s glocal (global-local) information environment admits so many more ways and means than denial and punishment to influence a target’s will and capability. Our language of strategy has not caught up to this reality. Therefore we should expand strategy beyond coercion theory to include cooperation and confrontation.
The combined effects strategy and concepts of influence (see Paper #41) described on this website is a start. We begin with 16 basic ways, means, and ends as concepts of influence (COI). Here are the sixteen (they include denial and punishment, but not just to compel or deter):
A strategist asks three basic questions, the “dimensions” of strategy, and applies them to the “elements” of strategy–ends, ways, and means: (1) cause and/or prevent? (2) cooperate and/or confront? (3) psychological and/or physical?
We don’t speak this language, but adversaries certainly behave as if they do. Consider the Russian Federation under President Vladimir Putin.
I analyzed 15 cases of hybrid competition (from cooperation to confrontation) between 2008 to the present day using this language. Here are the effects and concepts of influence:
(1) Ukraine, 2004: coerced compellence via denial and intimidation
(2) Estonia, 2007: compellent coercion via intimidation and punishment
(3) Lithuania, 2008: compellent coercion and inducement via intimidation, punishment, and exercise
(4) US, 2008: defensive deterrence via denial, neutralization, and intimidation
(5) Georgia: coercive compellence via denial and intimidation
(6) Kyrgyzstan, 2009: inducible compellence via exercise and intimidation
(7) Kazakhstan, 2009: inducible compellence via demonstration and intimidation
(8) Ukraine, 2013: inducible compellent deterrence via exercise, intimidation, and neutralization
(9) Ukraine, 2014: coercive deterrence and compellence via denial, neutralization, and intimidation
(10) Netherlands, 2015: deterrence and compellent coercion via denial and neutralization
(11) Syria, 2015: coercive deterrence and compellent inducement via denial, neutralization, intimidation, and exercise
(12) US, 2016: inducible compellence and coercion via demonstration and neutralization
(13) Switzerland and NATO, 2018: coercive deterrence via intimidation and denial
(14) Global, 2020: coercive deterrence and defense via denial and neutralization
(15) Germany, 2021: persuasive deterrence and compellence via enhancement and neutralization
In this sample, the prevailing pattern is to combine compellence, coercion, and deterrence. Those are the most frequent effects—compellence (11), coercion (10), and deterrence (8).
Russian methods (defined here as ways and means) in order of frequency are intimidation (11), denial (8), neutralization (8), exercise (5), punishment (2), and demonstration (2).
None of the cases combine denial with punishment.
The most often frequent combination of effects is compellence and coercion (8), followed by compellence and deterrence (5), then coercion and deterrence (4).
We need to embrace this complexity now before an authoritarian AI goes rogue, disables a nuclear command and control system, or creates other forms of human-uncontrolled vertical or horizontal escalation.
Competitive strategy is becoming an advanced survival skill. AI writes code and learns to iterate rapid solutions that humans cannot. AI has no moral agency. It’s like an unaccountable authoritarian actor that competes by sowing chaos, but much worse. AI capabilities are inherently dynamic.
Can we expect Putin’s Russia to abide by restraints on AI-enabled operations? The ability of AI to generate novel, unexpected results should force us to negotiate controls and limitations as in the cold war period.
As for China, its single-Party autocracy refuses to enter negotiations until it has achieved military equivalence. The mutual dread of uncontrollable AI could force China into negotiations prior to the point of nuclear mutual assisted destruction.
Those considerations are why strategy in an AI-enabled IE needs to integrate diplomatic, informational, military, economic, and social (DIMES) concepts of influence. Our “when-deterrence-fails” language of strategy puts diplomacy in the lead by default. A combined effects framework considers all types of approaches and effects in terms of the three dimensions of strategy. There is no default type of cooperation or confrontation.
AI is already a competitive aspect of whole-of-government and private sector strategies. Besides military and social applications, AI governance is part of state diplomacy and AI trading technology mitigates risk.
The following summarizes my four proposed requirements for effective strategy in an AI IE.
1. Speaking a language of strategy that’s more holistic, agile, and asymmetric than brute force or coercion.
2. Possessing the will and capability for all-effects competition and warfare, not just all-domain warfare.
3. Establishing a profession of effects beyond the profession of arms.
4. Implementing DIMES-wide combined effects and concepts of influence.