We are talking about Artificial Intelligence (AI). Everyone is talking about AI. Eh, AI! Or simply: AI.
But what is AI, really? How present is it in our lives—too much, too little? Where are we heading? Is it something to be feared, or embraced? Have our phones already taken control over us? Man, or machine?
Not long ago, I passed a shop window with several laptops on display. It reminded me of an older movie scene: a laptop switches on, a human face appears on the screen—eyes, nose, mouth—and then it starts to speak, trying to sell itself to me. After a quick mental sequence, I thought: “This is it, we’re not far off!” A computer initiating a negotiation process with the purpose of concluding a sales contract is precisely an act of negotiating with AI.
But what do we believe—or rather, what do we like to believe—about ourselves? That we are in control, proud guardians of free will (I almost said, like the “free Dacians”), and that we could never be—quote—“controlled.” But is that really so?
Free Will: A Philosophical Dilemma
At home, I opened Plato’s Dialogues with Euthyphro:
“And what is the matter about which, if we differ, we shall become enemies and be at enmity with each other? Perhaps you do not see at once; I will tell you. They are the just and unjust, the beautiful and the ugly, the good and the evil. Are these not the things about which, when we differ, and cannot come to a satisfactory decision, we quarrel and become enemies, you and I and all men?”
And then comes the passage that introduces our theme: “It is not impossible that the same thing may be loved by the gods and hated by them.” “One will differ from another, and no one will judge the same as another concerning the same matters.” (I know—you expected the Allegory of the Cave when I mentioned Plato!)
From here, the dilemma grows like Făt-Frumos in Romanian fairy tales: in three days and three nights. What determines free will? Can good faith in negotiation be controlled or monitored, given that a computer program may use a spectrum of information far greater than a human? If the mental operations of forming consent are adjustable—even slightly—so that a contract results, do we accept that this legal condition is absent? And if I were defining the algorithms, would I resist the temptation to equip the program with the necessary tools to achieve a well-determined goal?
The answer to the last question is certain: no.
Free Will: Who Has It?
Free will is the human ability to choose a course of action or make a decision through autonomous thinking, generated by internal factors and shaped by external ones, without imposed limitations. It also means “thinking about thinking”, extracorporeal perception, the individual’s relation to the whole and vice versa, acceptance of infinity, enlightenment, and illumination.
By contrast, AI means mathematics—algorithms that analyze data to produce results. Take chess: given a position, the program generates possible moves, tests them, assigns scores, and chooses the best one—or the first one “good enough.”
As John Searle’s Chinese Room thought experiment shows, intelligence is not the same as intelligent behavior.
Programs are built with conditions (“ifs”). When met, they trigger effects.
Thus, AI cannot make decisions beyond algorithms, nor illogical ones. Humans can.
One might be tempted to attribute AI a semblance of free will, since it evaluates alternatives. But because mathematics is finite and limited, while free will is not, the conclusion is clear: AI does not have free will.
Humans, on the other hand, compare consequences—not just scores.
Point for humans.
Freedom to Contract
Parties have the freedom to initiate, conduct, and break off negotiations, as stated in Article 1183(1) of the Civil Code.
A contract is the legal act by which individuals pursue personal interests. Contractual freedom implies the ability to negotiate, compare, and act according to expectations and goals. AI may guide—or even shape—human interests.
Here, the principle of contractual freedom is tested. By design, AI will use static and dynamic techniques to reach predetermined outcomes. Humans may find themselves bound by contracts against their own interest. Think about the last time you bought something online that you didn’t really need.
Even against a skilled negotiator, AI will remain inert compared to humans—it cannot respond to external emotional triggers such as pity, fear, greed, temptation, duty, curiosity, or pride. These instruments, always present in negotiation, create imbalance. The manipulative potential of AI could guide the human interlocutor’s perception.
Thus, the human mind’s ability to decide freely—based solely on self-controlled factors—comes into question, because it is permeable to external inducement. It is difficult to determine how much of a choice is true free will, and how much is programmed stimulus.
Score: 1–1.
Formation of Consent
A contract is an agreement, usually negotiated, built on rational and psychological elements of legal consciousness, grounded in the principle of autonomy.
But an offer to contract requires sinalagma—the meeting of wills (duorum vel plurium in idem placitum consensus). Since computer programs rely on conditions subject to validation, can ethical principles generating genuine consent exist in AI?
True ethical and moral decisions require understanding concepts of good and evil, right and wrong, reasonable and unreasonable. AI lacks psychological experience.
Thus, AI cannot meet the legal requirement under Article 1204 of the Civil Code: consent must be free and informed.
Score: 2–1 for humans.
Conclusion
AI should not be excluded from negotiations. It may be a useful tool in low-value, high-volume contracts or standard-form agreements. In such cases, the definition of when a contract is concluded—Article 1182 Civil Code, at the moment offer meets acceptance—remains valid.
Yet the use of AI raises questions of liability. Does it belong to the contracting party who uses AI as an instrumentum? Or to the programmers who wrote the source code? This is a separate issue.
Ultimately, biological intelligence has distinct advantages over artificial intelligence, and most transactions can—and will—be negotiated only by humans.
Av. Paul Vargan
Managing Partner, VARGAN & VARGAN S.C.A.
Coordinator of the MyJustice – Digital Legal Solutions® project
Ideas presented at the “Legal Preview of Artificial Intelligence” conference organized by juridice.ro on December 7, 2022.
