by dr Andrzej Jakubiec - adwokat, mediatorCOUPLED CONFIRMATION BIAS – MY CONCEPT
In this article, I present my original proposal of the Coupled Confirmation Bias (CCB). It is a conceptual framework designed to analyze the escalation of conflicts in situations where both parties to a dispute independently rely on AI systems to interpret the conflict and to justify their own positions. Unlike classical confirmation bias, which operates at the level of the individual, CCB describes a systemic mechanism. In this mechanism, mutually reinforcing narratives lead to a progressive narrowing of the negotiation space and an increased likelihood of escalation. The article identifies the conditions under which this mechanism is activated, its typical consequences, and the limits of its applicability. The purpose of this text is not to present a fully validated theory. Its purpose is to formulate a hypothesis regarding the existence of a repeatable mechanism observed in contemporary conflicts. What these conflicts share is that AI models are used as analytical tools supporting decision-making.
1. Introduction: When Rational Tools Amplify Irrational Outcomes
Where does the Coupled Confirmation Bias come from? People increasingly rely on AI. They use it to assess existing relationships both before a conflict emerges or becomes consciously recognized, and during the conflict itself. AI is used to assess risk, develop strategies, and justify proposed actions.
AI models are also increasingly used to analyze statements and non-verbal actions of the opposing party. AI systems are commonly perceived as authoritative, and their recommendations as neutral, objective, and free from emotional involvement. Paradoxically, however, as I have observed, the use of language models often correlates with faster conflict escalation, rigidification of positions, and premature breakdown of negotiations. This does not appear to be a coincidence.
In this article, I argue that these phenomena cannot be sufficiently explained solely by individual cognitive biases. They cannot be explained either by simple confirmation bias or by the fundamental attribution error.
Instead, I assume the existence of a systemic mechanism: the Coupled Confirmation Bias (CCB). It emerges in situations where parties to a conflict independently use AI tools as a source of authoritative interpretationof the dispute.
This interpretation may concern external factors, factual actions, and declarations of the opposing party. It may also concern conjectures about the opponent’s true intentions, acceptable risk, costs they are willing to bear, and their actual so-called “red lines”, meaning critical points they will not allow to be crossed under any circumstances.
Naturally, not all of these elements must be analyzed using AI. They also do not need to be analyzed at the same time. Importantly, each party may analyze a different element or set of elements.
Finally, each party may use different types of models, with different levels of technical sophistication and different degrees of integrated interaction with the user.
The Coupled Confirmation Bias (CCB) produces stronger effects the more often parties analyze signals coming from the opposing side when the content, form, or timing of those signals has previously been influenced by AI-assisted work conducted by the other party.
We are therefore not dealing with a one-time error. We are dealing with a spiral of errors, where each previous step becomes the fuel for the next one and potentially amplifies it.
This can be illustrated by a ping-pong exchange in which, after each hit, the ball gains energy equal to the sum of its previous kinetic energy and the energy of the new strike.
I also assume that the Coupled Confirmation Bias (CCB) will manifest in a similar manner across different levels of conflict. This includes conflicts between individuals, between groups, and between states or blocs of states. This conclusion follows from general research on the nature of conflict.
I wish to emphasize that I do not believe AI’s influence on conflict escalation is deterministic in nature. Rather, I assume the existence of a tendency—an influence. Moreover, awareness of this mechanism, as I understand it, can paradoxically lead to a halt in escalation, once the parties realize they are under its influence.
2. Current State of Research
The equilibrium mechanism between parties was described in various works by J. Nash, and conflict strategy by T. Schelling (The Strategy of Conflict).
The conflict structure adopted here is drawn from Christopher Moore’s book The Mediation Process: Practical Strategies for Resolving Conflict.
General cognitive biases have long been known to psychology.
For the development of the Coupled Confirmation Bias (CCB) concept, the starting point consists of works describing classical confirmation bias.
This bias refers to the tendency toward selective and subjective perception of data in order to confirm an already adopted thesis.
A well-known example is Daniel Kahneman’s Thinking, Fast and Slow.
This mechanism has also been identified at the level of human–AI interaction and described in detail, among others, in articles by M. Glickman and T. Sharot – https://pmc.ncbi.nlm.nih.gov/articles/PMC11860214/?, Yuxin Liu and Adam Moore –https://pubmed.ncbi.nlm.nih.gov/40448478/ , as well as L. Celar and Ruth M. J. Byrne –https://pubmed.ncbi.nlm.nih.gov/36964302/.
It is also necessary to mention the article by Ben Wang and Jiqun Liu, Cognitively Biased Users Interacting with Algorithmically Biased Results in Whole-Session Search on Debated Topics – https://dl.acm.org/doi/10.1145/3664190.3672520.
These authors point to the role of individual factors in susceptibility to cognitive biases arising in interaction with artificial intelligence.
From these works we learn about the feedback loop mechanism. It does not merely involve the presence of confirmation bias. It goes one step further and leads to reinforcement of the initial belief or prejudice, resulting in tunnel thinking.
However, in the works known to me, researchers have not examined dispute situations in which both sides use language models in the manner described above. The Coupled Confirmation Bias (CCB) is not the sum of two parties’ cognitive biases. It is a distinct phenomenon, a new quality that leads to escalation spirals in a manner qualitatively different from two independent errors.
3. From Individual Bias to Systemic Escalation
Classical confirmation bias describes an individual’s tendency to selectively search for, interpret, and remember information in ways that confirm prior beliefs.
Although this phenomenon is well documented, it is insufficient to explain situations in which both parties to a conflict, despite having access to initially similar data and using ostensibly neutral analytical tools, become increasingly entrenched in their own beliefs and prejudices.
In contemporary disputes, AI systems increasingly function as external legitimizers of interpretation, rather than merely computational tools. When each party uses such systems independently, confirmation bias ceases to be merely an individual cognitive tendency and begins to function as a mutually coupled dynamic.
4. Definition of the Coupled Confirmation Bias
The Coupled Confirmation Bias (CCB) is a conflict escalation mechanism in which two or more parties to a dispute, relying on external interpretive systems perceived as epistemically privileged, mutually legitimize their own narratives. This mechanism is recursive. Actions taken on the basis of such legitimization subsequently become input data for further analysis on the opposing side. This leads to a coupled interpretive spiral and a gradual narrowing of the negotiation space.
The constitutive feature of CCB is not merely the presence of biased reasoning. It is the dynamic coupling of interpretive loops between actors. In this model, actions of one party shaped by AI analysis become direct input for the system used by the other party. This creates a closed circuit in which each successive interaction does not bring the parties closer to consensus. Instead, it provides “objective” material for deepening the original prejudices.
5. Hypothesis
H1:
In bilateral or multilateral conflicts where parties independently use AI models to interpret the dispute and justify their own positions, the probability of escalation and negotiation breakdown is higher than in structurally similar conflicts in which such systems are not used.
H1a:
In interpretation-based conflicts, symmetrical access to information increases the risk of escalation more than informational asymmetry.
This is because it eliminates the possibility of explaining disagreement through lack of knowledge and shifts the conflict to the level of intentions and alleged rationality.
5.1.
I am aware that AI models may be used by each party at different times, to different extents, for different purposes, and in different ways. They may begin using AI simultaneously or at different moments. One party may stop or limit its use earlier, while the other continues. One party may analyze only external factors, another party declarations, and at another time attempt to infer the opponent’s intentions using AI. One party may use AI for analysis, another for emotional support. Finally, parties may use different models and interact with them differently, including by providing more or less manipulated input data. As demonstrated by the aforementioned research of Ben Wang and Jiqun Liu, individual factors also influence susceptibility to AI suggestions and outputs.
All these factors must be taken into account. I recognize that resulting differences may cause the Coupled Confirmation Bias (CCB) not to emerge or to dissipate during the conflict. At this point, my thesis concerns situations in which both parties use AI in a relatively symmetrical manner. The impact of asymmetry on CCB requires further research.
5.2.
Research has repeatedly shown that systems composed of three actors are significantly less stable than those involving two or four actors. I believe that the role of AI in accelerating escalation will be particularly visible in three-actor configurations. Further research will need to determine differences in outcomes when, in systems of three or more participants, only some of them use AI.
5.3.
I wish to emphasize clearly that AI does not create or escalate conflict by itself. However, it has a powerful influence on user perception, interpretation, and ultimately decision-making. Actions taken as a result of such decisions are then interpreted in an analogous manner by the opposing party.
5.4.
The Coupled Confirmation Bias (CCB) is also not an example of so-called echo chambers limited to two or a small number of participants. Echo chambers are inherently static. CCB is dynamic, because each subsequent interaction changes the behavior of the other party.
6. Falsifiability of the Coupled Confirmation Bias (CCB)
In The Logic of Scientific Discovery, Karl Popper introduced falsifiability as a necessary condition for recognizing a theory as scientific.
It is therefore necessary to specify which factors condition the emergence of the described mechanism, which contribute to its deactivation, and above all, what would falsify this theory.
6.1. Conditions for Activation
The Coupled Confirmation Bias typically emerges when the following conditions are jointly met:
Symmetrical legitimization of narratives
Each party has tools or advisors confirming its interpretation as rational and justified.
Absence of a shared epistemic authority
There is no institution, mediator, or procedure recognized by all parties as a final arbiter.
High cognitive cost of changing position
Changing position would undermine earlier “rational” decisions supported by AI analysis.
Presence of an apparently neutral third actor
AI systems, perceived as objective and interest-free, reinforce the legitimization of each narrative.
Paradoxically, this facilitates their functional contribution to confirmation bias on both sides and, consequently, to the emergence of the Coupled Confirmation Bias (CCB).
6.2. Limits of Applicability: When CCB Does Not Operate
The Coupled Confirmation Bias is not universal. The mechanism weakens or does not occur when:
– A commonly recognized factual arbiter exists.
– Only one party uses AI tools, breaking coupling symmetry.
– The stakes of the conflict are low or reversible.
– AI is used exclusively for information, not interpretation.
– Parties possess high metacognitive competence and actively counteract their own biases.
– The dispute concerns clearly measurable parameters rather than interpretations of intent or attribution of blame.
– Significant asymmetry arises in AI usage regarding time, scope, purpose, or method.The CCB mechanism is also weakened when parties are mutually aware of using similar interpretive tools and are capable of metacognitive reflection on their influence. Disclosure of AI use may itself become a de-escalatory factor.
These limits distinguish CCB from general theories of conflict escalation.
6.3. What Would Falsify This Theory?
The hypothesis would be falsified by observing the opposite effect: a mitigating or de-escalatory influence of AI models on conflict dynamics.
Such an effect might appear if an AI system suddenly altered its narrative after identifying a feedback loop and explicitly informed the user of its existence.
I have not observed anything of this kind to date.
The hypothesis could also be falsified if earlier findings on feedback loops and tunnel thinking proved incorrect, or if AI evolution fundamentally altered interaction principles.
At present, I am not aware of evidence supporting such claims.
Falsification would also occur if escalation in observed cases were shown to result from factors other than cognitive biases arising from AI interaction.
Finally, the hypothesis would be falsified if conflict dynamics were identical regardless of whether participants used AI.
Current observations contradict this.
7. Systemic Effects: The Dynamics of Recursive Escalation
Activation of the CCB mechanism shifts conflict from contested interests to closed interpretive loops.
This entails the following systemic consequences:
Autocatalytic escalation
Due to its recursive nature, every de-escalatory communication attempt is filtered through the opposing party’s AI. If the system interprets the other side as disloyal, even goodwill gestures are framed as strategic manipulation, paradoxically reinforcing escalation.
Ambiguity collapse
In classical disputes, uncertainty about intentions leaves room for interpretation. In CCB, AI “closes” interpretations by granting them analytical certainty. Parties stop discussing facts and operate instead on finalized analytical outputs—judgments about the opponent’s intentions.
The legitimacy trap
Because each party possesses “objective” confirmation of its position from a subjectively epistemically privileged system, compromise becomes framed as irrational or logically flawed.
Erosion of shared reality
Recursive layering of analyses causes parties to stop responding to real actions and instead react to how their own AI models predict the opponent’s AI interpretations. The conflict detaches from reality and moves into a model-to-model interaction space.
Position inertia
AI-shaped narratives exhibit strong resistance to change (cognitive inertia). Challenging conclusions generated by advanced analytical models would require decision-makers to admit a fundamental error in tool selection. This raises both psychological and “political” costs of de-escalation.
Importantly, this escalation occurs without bad faith and often without conscious strategic intent.
8. AI as a Quasi-Third Actor?
Although artificial intelligence lacks agency and its own interests, its functional role in the CCB mechanism cannot be overlooked. By providing authoritative legitimization while simultaneously lacking accountability, AI systems influence the dynamics of a dispute. They affect the subjective assessment of escalation costs and stabilize mutually exclusive narratives. To be clear—I am not suggesting that AI’s role grants it the status of a party in an ontological sense. Rather, it acts as a mirror that actively reinforces primary beliefs. However, it is not a passive mirror, but an active one—invested with trust and strengthening convictions.
9. Conclusion
The Coupled Confirmation Bias (CCB) provides a conceptual framework for understanding why introducing ostensibly rational tools may accelerate conflict escalation. The proposed concept does not claim universality or empirical finality. It is a hypothesis of a mechanism observed in practice, inviting further criticism, testing, and refinement.
Understanding CCB is relevant not only for lawyers, mediators, and conflict managers. It is also significant for designers and users of AI systems in confrontational contexts.
10. Author’s Methodological Note
This article presents my original analytical and conceptual framework based on patterns I have observed in conflicts. These observations stem from my own legal practice in the second half of 2025. The article does not constitute an empirically validated theory. It is a hypothesis. Its purpose is to describe a phenomenon and indicate directions for further analysis. In developing this work, I used language models to critically evaluate my own theses and to identify weaknesses in my reasoning. AI assistance was critical rather than generative. It served to identify potential vulnerabilities in the argumentation.
You can read also in Polish:
by dr Andrzej Jakubiec - adwokat, mediatorHOW WILL AI IMPACT DISPUTE DYNAMICS IN 2026?
Recently, I wrote about AI’s impact on escalating family and business conflicts. I described the coupled confirmation bias. I showed how it leads to tunnel vision. This creates a micro-scale version of the security dilemma. Today, I share reflections on how Large Language Models (LLMs) will change dispute dynamics in 2026. Current trends will likely not reverse on their own. Instead, everything points to their significant intensification.
How AI Influences Dispute Dynamics?
AI directly shapes how conflicts evolve. I will not repeat my previous articles here. Instead, I am providing links to the most important ones. They contain links to the latest scientific publications and my other texts. This article does not provide legal advice for Poland. It explains general conflict dynamics.
I wrote about people using AI to diagnose and solve important legal problems:
https://jakubieciwspolnicy.pl/ai-zastapi-prawnikow/
In that same text, I explained why such analysis is insufficient. It is often incomplete and requires legal verification.
You can read about how lawyers and clients use AI:
https://jakubieciwspolnicy.pl/korzystanie-z-ai-przez-prawnikow-i-klientow-podstawowe-problemy/.
However, my most important article is available at this link in polish version https://jakubieciwspolnicy.pl/ai-i-myslenie-tunelowe-w-sporach-miedzy-wspolnikami/ and the english one: https://jakubieciwspolnicy.pl/en/ai-and-tunnel-vision-in-shareholder-disputes/ . I describe the universal rules of the conflict, so don’t hesitate to read it. It is not any analysis of the polish law.
I described the mechanism of coupled confirmation bias in detail there. I showed how it leads to dangerous tunnel vision.
The greatest risk arises when both parties use AI models to interpret each other’s behaviors. This can lead to rapid, uncontrolled conflict escalation.
Read more about the feedback loop here: https://pmc.ncbi.nlm.nih.gov/articles/PMC11860214/ This article by M. Glickman and T. Sharot discusses how AI feedback loops change human judgment.
I also recommend the text regarding causal explanations in AI by L. Celar and R.M.J. Byrne: https://pubmed.ncbi.nlm.nih.gov/36964302/
Finally, I suggest the article by Liu Y and Moore A. It covers intuitive judgments regarding AI and moral transgressions. It is valuable for understanding social perceptions: https://pubmed.ncbi.nlm.nih.gov/40448478/
The Impact of AI on Conflict Intensity in 2026
There are no rational reasons to expect a reversal of current trends. We see growing AI accessibility and rising trust in AI. Furthermore, the computational capabilities of language models continue to expand.
I assume a group of people will become emotionally dependent on their “relationships” with AI. Withdrawn or lonely individuals will be particularly vulnerable. Those not used to critical thinking face the highest risk. A lack of critical assessment makes the AI model a “moral authority” rather than an information source. If you doubt human emotional bonds with machines, remember the Tamagotchi craze.
I believe AI’s role in shaping human emotional attitudes will grow. This will directly affect how we perceive current relationships. We will observe increased atomization and polarization of individuals and groups. Atomization occurs because AI companionship may seem more attractive than human contact. Already weak social ties will weaken further. Polarization will result from frequent human’s consultations with AI on sensitive matters like family or business partnerships. In a dispute, users might not seek a solution. Instead, they may seek validation for their own narrative.
These mechanisms will become increasingly common. Public awareness may not keep pace with AI’s real-life impact.
Can AI Be Helpful in Resolving Disputes?
Yes! Despite the risks, AI will also have strong positive aspects. AI will serve in prevention. It helps predict potential conflict areas and secure them early. Language models are helpful here. They easily create multiple scenarios and highlight potential risks. They will certainly function as an effective early warning system.
Generative AI will also help find rational solutions to existing disputes. Its ability to multiply potential outcomes is amazing. After describing a deadlock, the chat may suggest a solution we missed. However, we must always critically evaluate these suggestions. We must predict their long-term consequences across many levels.
In every case, the role of AI remains auxiliary. It builds variants well but struggles to analyze legal and emotional consequences.
At this point, reference should also be made to the research of Marco Giacalone, who points to the significant potential of language models in dispute resolution: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5083207. The author writes that ‘The integration of generative AI reduces costs and allows legal practitioners to focus on complex issues, strategic planning, and client interaction.’ According to the author, AI will not replace humans, but will be a very valuable tool in their hands. I agree with this approach.
AI Influences Dynamics but is Neither Good Nor Evil
It makes no sense to ask if AI is “good” or “bad.” We should study its impact within specific contexts. Our species created AI. Trying to put the “genie back in the bottle” is impossible and pointless. Let’s objectively study its impact on our thought patterns and relationships. We should draw logical conclusions from these observations. Use this technology to benefit yourself and others. I believe this is possible. However, do not let AI replace human friendship and intimacy. Do not allow law and health to fall under unreflective AI influence.
We are dealing with a powerful new tool. It is a new source of influence on our psyche. Humanity has never known a tool this powerful. Yet, every great discovery brings both opportunities and threats. Ultimately, the outcome depends entirely on us.
Unlike nuclear energy, almost everyone now has direct access to AI. AI will amplify what is already hidden and strong within us.
Invitation to Cooperation
Have you noticed AI suggestions influencing your private or business relationships? Do you see this impact on yourself?
Perhaps you notice the other party becoming more radical lately?
I invite you to discuss this in the comments or contact our law firm. We help clients build strategies and resolve disputes optimally.
We consider many factors: law, psychology, communication, and technology. This is our strength.
📩 kancelaria@jakubieciwspolnicy.pl
📞 +48 536 270 935
by dr Andrzej Jakubiec - adwokat, mediatorAI in Family Disputes: My last experiences and A New Reality
In my recent posts, I discussed AI’s role in shareholder disputes. Today, I want to focus on a very sensitive area. I am referring to family law matters. I will describe cases from my own practice. These events occurred recently, in December 2025. As a conflict specialist, I find these examples particularly interesting. I am observing AI’s growing role in creating and escalating disputes. This trend is especially visible in family-related cases.
AI in Family Cases Matters and Its Impact on Human Behavior
AI in family cases is no longer a science – fiction. It is a reality. I increasingly notice AI’s strong influence on family disputes. In this post, I will share cases from recent weeks.
I have changed all details to protect the privacy of those involved. No one can be identified from these descriptions.
I spoke with each person several times, both in person and online. I am describing a very new phenomenon.
Its consequences for family life and mental well-being are still unknown. This applies to individuals, families, and entire communities.
I avoid making judgments in this text. I prefer to leave the evaluation to you.
CASE ONE: CHILD CONTACT AFTER DIVORCE
A client recently contacted me after his divorce. I did not represent him in those proceedings. He is a successful, intelligent man in his thirties.
He needed urgent advice regarding child contact. Two weeks earlier, he agreed to a specific schedule. It included set weekdays and every second weekend.
Crucially, the former couple still lived together. His ex-wife planned to go out for the weekend. She expected him to care for the children as agreed.
The client asked an AI model about his obligations. He wanted to know if the schedule was a duty or just a right. He viewed it as a “safeguard” against his ex-wife.
The AI confirmed his incorrect belief. It stated he had no obligation to care for the children. This led to a violent argument between the parents.
Lawyers should explain these basic rules to their clients. However, the AI provided incorrect legal advice. It reinforced the client’s bias.
This is a perfect example of a coupled confirmation bias. The user described only a fragment of reality. The AI then validated his mistaken views. This led to immediate confrontation and escalation.
CASE TWO: AI AS A HIRED PSYCHOANALYST
This client is a highly educated, high-earning woman in her thirties. While preparing for her case, she asked an AI to psychological profile of her husband.
She described him entirely from her own perspective. The AI replied that it was not a psychologist. However, it still offered to help.
The model assigned numerous disorders to the husband. It labeled him as emotionally immature, narcissistic, and psychopathic.
Based on this AI chat, the client lost all trust. She decided it was unsafe to leave the child with him and was ready form cut of fathers contacts with their baby.
I suggested she consult a real psychologist. She refused, claiming the AI had already answered everything. She was now ready to block all contact between father and son.
The problem is that she treated a chat as a clinical diagnosis. It was based on a one-sided description. AI cannot replace a licensed psychiatrist or psychologist.
A professional must examine the patient and use proper tests. They cannot rely solely on the report of an involved party. I declined to take this case.
CASE THREE: AI AS A CONVERSATION ANALYST
Two parents came to me for mediation. Initially, their cooperation was good. They communicated mostly via long WhatsApp messages.
Soon, they became deeply suspicious of each other. They looked for hidden motives and secret plans in every text. Both believed the other was using the children as tools.
It turned out that both parents were using AI. They pasted received messages into the chat for detailed analysis. The AI then helped them draft “strategic” replies.
The other party would then analyze those AI-generated replies using their own model. This created a spiral of suspicion. Both parties began to lose touch with reality. They were ready for radical, harmful steps.
AI in Family Matters – Conclusions
These three examples from my practice show that we tend to:
- Search for confirmation of our initial assumptions.
- Strengthen our beliefs after receiving such confirmation.
- Prefer AI as a source of validation because it is faster, cheaper, and “politer.”
AI often carries an aura of omniscience. This makes it seem more attractive than a lawyer or a psychologist.
This leads to tunnel vision. People become ready to escalate disputes quickly and without deep thought. Our prejudices grow stronger. We believe we have received confirmation from “reliable” technology.
The long-term effects of this fascination with AI are hard to predict. But the problem is not the technology itself. The issue is that many people cannot critically evaluate AI responses.
AI also intensifies the “fundamental attribution error.” I have written about this here: [LINK]
Clients often come to my office with ready-made “solutions.” They expect me to simply implement them. I have discussed this trend since early 2025 here: [LINK]
You can read more about tunnel vision and coupled confirmation bias here: [LINK]
Scientific Research: AI and Cognitive Biases
My observations are supported by scientific data. AI increasingly influences human beliefs, choices, and behaviors.
I highly recommend an article published last year: How human–AI feedback loops alter human perceptual, emotional and social judgements (Nature Human Behaviour). The authors show that AI validation strengthens our perceptions and social judgments. Read it here: https://www.nature.com/articles/s41562-024-02077-2
You should also read Yiran Du’s work: Confirmation Bias in Generative AI Chatbots. It analyzes confirmation bias mechanisms in AI models and the associated risks: https://arxiv.org/abs/2504.09343
Another important text covers tunnel vision: Bias in the Loop: How Humans Evaluate AI‑Generated Suggestions. Experiments prove that users accept wrong AI suggestions if they fit their prior beliefs: https://arxiv.org/pdf/2509.08514
Finally, here is an analysis from Stanford University. It examines AI “hallucinations” and their impact on decision-making: https://hai.stanford.edu/news/ai-trial-legal-models-hallucinate-1-out-6-or-more-benchmarking-queries
Contact Me
If you need a lawyer who specializes in dispute resolution—including family law—please reach out:
📩 kancelaria@jakubieciwspolnicy.pl 📞 536 270 935
I am here to help you!
by dr Andrzej Jakubiec - adwokat, mediatorAI and Tunnel Vision in Shareholder Disputes
What is the problem?
The Coupled Confirmation Bias is more and more often present in business disputes. I can see it even in my current practice. Sometimes, clients come to me with a ready-made plan. They have already consulted a language model about their company’s situation. They expect me to simply execute it. Conversations with them can be difficult. I feel a strong need to perform my job properly and investigate the situation. Some clients understand this. Others assume it is unnecessary because the case is already “assessed.” A few have even accused me of inflating costs. They claim AI has already done the work and provided a solution.
Confirmation bias clearly influences some clients’ attitudes. I manage to convince a portion of them. Others are surprised when I refuse to cooperate. Some are even outraged. One person accused me—before any substantive talk—that I don’t understand companies or negotiations. My ‘ignorance’ supposedly stemmed from my desire to read the articles of association. I also wanted to discuss the history of the partnership. Welcome to the AI era!
This article expands on thoughts I shared previously.
You can read that text first, but it is not mandatory. This article stands on its own.
Shareholder Disputes: Why Do We Seek Confirmation?
I have observed a natural tendency in shareholder disputes for years. People selectively choose facts that support their version of events. In psychology, this mechanism is called confirmation bias. We ignore information that contradicts our beliefs. Meanwhile, we overvalue evidence that confirms them.
Example: If we believe the Earth is flat, we interpret data to prove it. We ignore inconvenient facts or stretch others. Confirmation bias is not an accusation against anyone. It is a well-researched psychological phenomenon. It is good to be aware of it.
Psychologically, this is a defense mechanism. Its goal is to maintain cognitive consistency and reduce emotional tension. Intellectual anxiety does not serve most of us. We want to eliminate it.
In disputes, every party presents a “favorable” version. They select facts and make convenient assumptions. They weave these assumptions into a factual narrative. This happens at every level: from playground fights to international conflicts. In shareholder disputes, confirmation bias reveals itself with full force.
Until now, clients often came to me with a ready “diagnosis” and “treatment plan.” They sought a lawyer who would accept it as truth.
You can read more here:
Coupled Confirmation Bias in AI Interaction
What is Coupled Confirmation Bias? It is a mechanism where our initial beliefs are reinforced by AI interaction. The pattern looks like this:
- The user formulates a thesis (e.g., a belief about a partner’s dishonesty).
- The AI model generates a response that matches this assumption. It lacks the full history of the partnership. It does not distinguish between facts, assumptions, and interpretations. It wants to be “helpful,” so it usually confirms the user’s thesis.
- The user treats AI as an authority. This authority is strengthened after receiving confirmation.
- A feedback loop forms. It leads to even stronger convictions and a deeper cognitive tunnel.
- New forces arise. AI reinforces beliefs. The user acts decisively, “ennobled” by the confirmation. The conflict escalates.
AI models often state they are not lawyers. However, this has a counterproductive effect. Many users then look for lawyers who will confirm the AI’s conclusions. They use this as the main criterion for evaluating a lawyer’s competence. They will search until they find one.
In practice, AI acts like a “magnifying glass.” It amplifies our starting positions, regardless of their truth. We remove everything from our field of vision that contradicts the original thesis.
You can read more about this problem here: Ben Wang, Jiqun Liu, Cognitively Biased Users Interacting with Algorithmically Biased Results in Whole-Session Search on Debated Topics, https://dl.acm.org/doi/10.1145/3664190.3672520
Why Does AI Naturally Confirm Our Beliefs?
Language models do not “think” independently. Their answers come from the statistical prediction of words based on the prompt’s context. If a question suggests a specific interpretation, the model creates a consistent narrative. The user sees this as objective confirmation.
AI responses are internally consistent. They rarely contradict themselves. However, they are not always externally consistent or grounded in reality. They may lack true legal knowledge (not to be confused with legal regulations). At first glance, they look like expert statements. Only a professional can spot the errors or omissions that invalidate the suggested direction.
We call it hypercustomization. You can read about it: https://journals.sagepub.com/doi/10.1177/23794607251347020
Practical Consequences of Coupled Confirmation Bias for Partners and Negotiations: The Feedback Loop
Coupled confirmation bias can deepen negotiation difficulties. AI confirms one side’s assumptions, making it harder to understand the other side. But what if both partners use AI and coupled confirmation bias affects them both?
Consider what happens when the other party describes their subjective perspective to their own AI model. The model confirms its point of view and reinforces it. The influence on the user’s actions is significant. The second partner observes this with growing suspicion. His reaction will be to prepare for an attack, which may be strictly defensive or pre-emptively offensive. But he wants to be sure that his interpretation is correct. What will he do? He turns to his own chatbot, subjectively describing what he sees. Guess what kind of answer he receives? Yes…
The situation can spiral out of control. It resembles a chess match between two cheaters using computers. This is no longer a normal game.
Every partner subjectively interprets the “opponent’s” behavior. They feed this subjectivity to an AI. The AI confirms the “wickedness” of the other side and suggests radical solutions. When clients come to my office in this spiral, rational arguments often fail to reach them.
We have a clear example of coupled confirmation bias on both sides, which provokes a Feedback Loop! It is a security dilemma on steroids.
Human-AI Confirmation Bias in Scientific Research
My observations are confirmed by scientific research. Studies show that human-AI interactions can reinforce prejudices and false beliefs.
Last year has been published an article: “How human–AI feedback loops alter human perceptual, emotional and social judgements” (Nature Human Behaviour). The authors showed that AI confirming human assumptions strengthens perceptions and social ratings. Here’s the link: https://www.nature.com/articles/s41562-024-02077-2?
I also recommend the paper by Yiran Du: Confirmation Bias in Generative AI Chatbots. It analyzes these mechanisms in AI models and discusses the risks of this coupling: https://arxiv.org/abs/2504.09343?
Another insightful text is Bias in the Loop: How Humans Evaluate AI-Generated Suggestions: The authors found that users accept wrong AI suggestions if they fit prior beliefs. However, effective collaboration depends on who evaluates the AI results and how the review process is organized. You can read it here: https://arxiv.org/pdf/2509.08514
This mechanism is at the forefront of AI research. It is a clear example of how AI affects specific areas of life, such as negotiations.
Summary
Coupled Confirmation Bias is not a new cognitive bias; rather, together with tunnel thinking, it creates a new conflict dynamic in which AI meaningfully influences both the perception and the escalation of the conflict.
Coupled Confirmation Bias proves that AI is not a neutral arbiter. Our subjective biases can be reinforced in a feedback loop. In shareholder disputes, this leads to bad decisions and conflict escalation.
A lawyer’s role has never been just to confirm a client’s ideas. Today, we must go further. We must help some people regain contact with reality.
Everything depends on us. AI offers great possibilities. We can instruct it to be critical of our ideas. It can play devil’s advocate. It can find gaps in our reasoning or suggest alternative explanations. AI is excellent at eliminating the fundamental attribution error in business disputes. I wrote about it here: https://jakubieciwspolnicy.pl/podstawowy-blad-atrybucji-w-pracy-adwokata/
When a client brings AI-generated advice, I don’t get offended. I talk to them. I almost always review the material. Sometimes, a suggested solution is interesting and fresh.
Usually, I gain the client’s trust by explaining how language models work. I show them solutions I can legally defend. Sometimes, a client returns after a few days and says they finally trust me—because the AI eventually agreed with my reasoning. In light of the above, it is a bittersweet success.
If you need a lawyer who handles negotiations and shareholder disputes, feel free to contact me: 📧 kancelaria@jakubieciwspolnicy.pl 📞 +48 536 270 935 I will be happy to help!
by dr Andrzej Jakubiec - adwokat, mediatorWhat Does J. Mearsheimer Teach Us About Shareholders Conflicts?
From the beginning of my work as a business attorney, I observed a striking pattern. In companies with three partners, disputes arose more frequently than in firms with two or four partners. For a long time, I treated this as an interesting curiosity. That changed after I read John Mearsheimer’s The Tragedy of Great Power Politics. Why three-partner companies are statistically more prone to disputes? International political theories offer surprising insights into modern business partnerships. Specifically, Mearsheimer’s theory of offensive realism may explain why three-person structures face inherent instability. By understanding these structural dynamics, entrepreneurs can defuse conflicts before they destroy company value.
International Relations Theory in Business
Is instability of three-partner companies real? Business analysis often draws on theories originally developed to explain great-power politics. John Mearsheimer’s seminal work, The Tragedy of Great Power Politics, presents a clear thesis. He says, that bipolar systems, consisting of two main actors, are significantly more stable than multipolar ones.
In a business context, this translates into the relative durability of companies with only two partners. In such a setup, each partner typically holds a clearly defined role and maintains a strong incentive to reach an agreement to ensure the venture’s survival.
The Third Partner as a Structural Risk Factor
Once a third partner is introduced, the situation becomes strategically complex. Over time, the third individual naturally begins to assess their relative position within the company hierarchy. Neutral events or private conversations may be misinterpreted as signs of a growing alignment between the other two, leading to a breakdown in trust.
As questions arise about the sense of further investing energy, trust, and capital, perceptions begin to shift. This change can radically alter internal loyalty and decision-making dynamics.
Can Mearsheimer’s Theory Explain Shareholder Conflicts?
But can a theory about nuclear powers really apply to a three-person tech startup? The short answer: surprisingly well. The issue is whether they can be used to analyze relationships between business partners. To begin, let us lay out his core arguments. These are:
Mearsheimer’s key arguments about why multipolar arrangements are unstable:
- Anarchy in the system – There is no central authority to enforce agreements or reassure participants. Each actor must rely on itself for security. This creates constant pressure to accumulate advantage and undermines stable trust.
- Offensive capabilities – All major actors possess power that can be used against others. This creates persistent incentives to increase power rather than rely on cooperation.
- Uncertainty about others’ intentions – Actors can never be fully certain about others’ motivations or future plans. As a result, they make worst-case assumptions that drive competitive behavior.
- Survival as the primary goal – Fear of being outcompeted or dominated shapes strategic behavior. Players seek relative advantage even at the cost of short-term cooperation.
- Power maximization behaviour – Actors do not stop seeking power once basic security is achieved. They continue accumulating power to prevent rivals from gaining advantage.
- Security competition becomes self-reinforcing – When one actor increases power, others respond in kind. This dynamic fuels escalation rather than long-term stability.
- Greater opportunities for miscalculation – Multipolar settings create shifting alliances and imbalances. These conditions increase the risk of misjudging intentions or capabilities, triggering conflict.
These assumptions and behavioral imperatives form the core of Mearsheimer’s offensive realism perspective.
Doesn’t this list sound eerily familiar?
When we consider interpersonal dynamics in small, closed settings like commercial partnerships, this list stops being abstract. Competing for relative advantage and mistrusting intentions often feels natural in tightly knit management groups. The same logic appears in global geopolitical systems.
If we consider the theoretical basis for such a conceptual transfer, it cannot be dismissed outright. At minimum, it functions as a legitimate intellectual exercise rather than a formal scientific claim. This framing is a thought experiment, not a strict scientific argument. Still, the parallels are striking and, I hope, clear.
The Security Dilemma at the Micro Level
This internal tension closely resembles the “security dilemma” applied on a micro scale. In this scenario, an increase in one party’s sense of security or influence triggers an instinctive fear in the others. Consequently, a partner may seek to weaken the rest of the group simply to strengthen their own position.
The desire to assume a destructive role—such as an informal judge or arbiter—often emerges. Historical precedents, such as the Roman triumvirates or the history of successor kingdoms following Charlemagne’s empire, confirm that these three-way arrangements are rarely durable.
In my mediation and advisory practice, conflicts in three-partner companies tend to escalate faster and more emotionally than in two-partner structures.
Strategic Consequences for the Company
Conflict escalation in a trio is usually faster and more destructive than in other configurations. As decision-making paralysis begins to erode the organization from within, partners often find themselves fighting each other more fiercely than they compete with the market.
Because you know your partners best, you are able to strike at their most sensitive points. Ultimately, vital energy is diverted into building internal coalitions instead of driving growth.
In real corporate disputes, this dynamic frequently leads to board paralysis, operational stagnation, or costly shareholder litigation.
Instability of three-partner companies. Why Three-Person Structures Are Worth Avoiding?
Companies with three partners are among the least stable business forms because the potential for unequal alliances—typically two versus one—is embedded in the very structure. Instead of creating synergy, the system often devolves into a continuous zero-sum game.
If you are planning a new company, a two-partner model is usually a safer strategic choice. Reading Mearsheimer is essential for any leader who wants to understand and mitigate these structural risks.
Scientific Foundations: Conflict Dynamics in Triads
To fully understand the risks, one must examine the mathematical and psychological frameworks governing three-party interactions.
Offensive Realism in Systems Without a Central Arbiter
A company’s management board often resembles an “anarchic” system. Without a dominant leader to enforce order, each actor must maximize their own power to ensure survival. In a triad, this logic produces a cycle of constant alliance reshuffling.
The Sociology of the Triad (Georg Simmel)
According to Georg Simmel, introducing a third person fundamentally changes group chemistry. Three distinct roles typically emerge: the mediator, the opportunist (tertius gaudens), and the dominator. This structural shift transforms simple cooperation into a complex struggle for influence.
The Physics of Intractable Conflicts
Modern sociophysical models suggest that three-group dynamics are inherently unstable. The human mind is wired to perceive coalition threats more acutely in triads, which often triggers defensive aggression and long-term instability.
Instability of three-partner companies. Key Sources on Three-Party Conflict Dynamics
The following materials provide a foundation for deeper risk analysis in three-partner companies:
- [PL] Jakubiec & Partners – Three Partners: When Conflict Is in the Air An analysis by Dr. Andrzej Jakubiec linking Mearsheimer’s theory with Polish company law.
- [US] John J. Mearsheimer – The Tragedy of Great Power Politics The foundation of offensive realism. Explains why actors in anarchic systems seek dominance.
- A study examining which coalition structures can form in three-player games. It analyzes the conditions under which players form two-player coalitions, a grand coalition, or act independently: https://www.mdpi.com/2073-4336/16/3/30?
- Balanced Weights and Three‑Sided Coalition Formation (MDPI Games) https://www.mdpi.com/2073-4336/1/2/159
- Dynamic Stability of Coalition Formation in Dynamic Games: https://www.sciencedirect.com/science/article/abs/pii/S0167637724000749?via%3Dihub
- Hedonic Games and Coalition Stability: https://en.wikipedia.org/wiki/Hedonic_game?
Most shareholder disputes do not begin with bad intentions, but with structural blind spots that could have been addressed years earlier.
Prior to entering a three-person partnership, ensure your agreement includes provisions to mitigate structural deadlock. I invite you to reach out for a consultation:
📩 kancelaria@jakubieciwspolnicy.pl
📞 536 270 935