Divide-and-Conquer Dynamics in AI-Driven Disempowerment

Abstract: AI companies are attempting to create AI systems that outperform humans at most economically valuable work. Current AI models are already automating away the livelihoods of some artists, actors, and writers. But there is infighting between those who prioritize current harms and future harms. We construct a game-theoretic model of conflict to study the causes and consequences of this disunity. Our model also helps explain why throughout history, stakeholders sharing a common threat have found it advantageous to unite against it, and why the common threat has in turn found it advantageous to divide and conquer. Under realistic parameter assumptions, our model makes several predictions that find preliminary corroboration in the historical-empirical record. First, current victims of AI-driven disempowerment need the future victims to realize that their interests are also under serious and imminent threat, so that future victims are incentivized to support current victims in solidarity. Second, the movement against AI-driven disempowerment can become more united, and thereby more likely to prevail, if members believe that their efforts will be successful as opposed to futile. Finally, the movement can better unite and prevail if its members are less myopic. Myopic members prioritize their future well-being less than their present well-being, and are thus disinclined to solidarily support current victims today at personal cost, even if this is necessary to counter the shared threat of AI-driven disempowerment.

• The paper's main contribution is a game-theoretic model that quantifies win probabilities using a generalized gambler's ruin framework.
• It demonstrates that complete unity within the pro-human movement (m=1) significantly increases the likelihood of overcoming AI-driven replacement.
• The study highlights the importance of coordinated legal, political, and social strategies to counteract divisive tactics in AI disempowerment.

Divide-and-Conquer Dynamics in AI-Driven Disempowerment: A Game-Theoretic Approach

The paper "Divide-and-Conquer Dynamics in AI-Driven Disempowerment," authored by Peter S. Park and Max Tegmark, presents a comprehensive game-theoretic model to analyze the conflicts arising from AI-induced disempowerment. Specifically, the study aims to elucidate the causes and consequences of disunity among those affected by AI-driven automation and suggests strategic insights for resistance movements confronting this technological shift.

Overview

The authors focus on the conflicts between two major groups: proponents of human roles augmented by AI (Movement 1, also termed the pro-human movement) and those advocating for unrestricted AI deployment aimed at replacing human roles (Movement 2, the AI replacement movement). By constructing a generalized gambler's ruin model, the study positions these movements within a dynamic system of conflict, exploring the probability of either group's victory based on several parameters, including unity, relative strength, randomness, and perceived threat levels.

Theoretical Model

The paper delineates its model by segmenting it into two main parts: the dynastic system and the game-theoretic analysis:

1. Dynamical Systems Approach: Using parameters such as the degree of unity (quantified by the number of parts $m$ for Movement 1 and $n$ for Movement 2) and the relative strength $s$, the model calculates the win probability of each movement. A robust finding is that for sufficiently large $s$, complete unity (i.e., $m=1$) provides the highest probability of victory for Movement 1.
2. Game-Theoretic Analysis: The analysis evaluates the conditions under which members of Movement 1, driven by varying perceptions of AI threat ($\hat{s}$), costs of unity ($c$), and benefits ($b$), choose to unite or defect. The model predicts potential strategic pitfalls: myopia, naivety, collaborationism, defeatism, and complacency, all of which undermine unified action against AI encroachment.

Empirical Implications

The study extends its theoretical findings to historical examples of military and corporate divide-and-conquer dynamics, critically examining AI industry leaders' statements that may promote such fragmentation among the opposition. Historical records of military strategy (e.g., Napoleon's campaigns) and corporate influence tactics (e.g., marketing of plastics) corroborate the model’s predictions. Furthermore, the paper suggests that leaders in the AI industry frequently make pronouncements that, intentionally or not, could amplify divisions within the pro-human movement.

Future Directions and Practical Considerations

On a practical level, the paper suggests that stakeholders combine efforts across various domains: legal (through strategic copyright lawsuits), political (advocating for AI regulation), and social (public awareness campaigns). A critical observation is the necessity for timely and cohesive action, as delay may favor the faster-moving AI deployment by corporations ("a race to get most people hooked on AI").

The authors call for strategic unity among differing stakeholders within the pro-human movement to enhance their collective efficacy, suggesting that synergistic cooperation can catalyze success in ongoing negotiations, legal challenges, and advocacy movements.

Conclusion

Overall, this research importantly highlights the impending socio-economic transformations driven by AI and offers a rigorous clarification of the dynamics at play between different human groups under technological transition pressure. The game-theoretic framework adopted serves as a valuable tool for anticipating reactions within these movements and fostering successful cooperation that might effectively counteract AI-driven disempowerment. Such insights could guide future strategic planning and policy formulation in AI ethics and regulation.