By – Dhuha Mir
Abstract
AI is being increasingly used in many autonomous commercial transactions which has raised questions regarding who is responsible when the built in AI system leads to contractual failures. Traditionally, it was humans who used to contract with each other and were responsible for any contractual failures. But now since AI systems are being used to carry out the contracts with less or no human intervention, it has become challenging on how to tackle the question of liability where AI is the responsible party.
Thus the article examines whether the already existing Indian Contract Act 1872 can allocate the liability in cases of transactions which involve autonomous AI systems. It argues that the AI systems should not be treated as individual units. Rather the liability for faults of the AI should be placed upon the humans that derive benefit from it. Even though the existing legal doctrines can address breaches by AI to a certain extent there is still a need for a clear framework to be established and necessity of judicial interference so that accountability is ensured.
Introduction
In today’s world AI systems are increasingly being used in various commercial activities which were earlier performed by humans. Now various businesses have adopted automated systems in order to negotiate prices, manage supply chains, process orders and to carry out transactions. As these systems have evolved, they can now function with little or no human intervention. The adoption of AI-assisted contract management platforms has changed the way that the commercial contracts are now negotiated and performed which has led to several legal challenges.
Contract law is based upon the assumption that the contractual obligations arise from human intention and are performed by humans themselves. Hence any breach is also a fault of these human actors and are held responsible for them. However, when AI systems make these decisions which result in a breach, it becomes very difficult to ascertain who is to be held liable. Thus, the question arises whether the existing contract law is capable of addressing such breach and also preserving the traditional concepts of consent, agency and liability in cases of AI driven transactions?
Since there is an absence of a legal framework that can tackle the issues and breaches due to AI contracting there is a need for legal intervention and reconsideration.
Human Nature of Contract Law
The Indian Contract Act 1872 was drafted in a period where only human actors were considered to be eligible for entering into contracts. The essential elements of a contract that is offer, acceptance, free consent and intention to create legal relations all presume that the parties to a contract are capable of understanding and voluntarily entering into contractual obligations.
For a contract to be valid it is essential that the parties agree to the same thing in the same sense. Concepts of fraud, misrepresentation, coercion and mistake all exist because the parties are assumed to be human and possess human judgement. These concepts cannot be applied in case of AI transactions as these concepts were built around humans and their decision making capabilities. Since AI systems lack the human emotions of consciousness, intention or legal personality it becomes challenging for these doctrines application.
Similarly, even the concept of agency is based on the fact that the contract is formed between two legal persons. An agent under section 182 of the ICA is said to have acted on the behalf of a principal and the acts of an agent are attributed to the principal. Thus, even though it is the agent who has acted the responsibility is ultimately traced back to a human or juristic person which is the principal. This concept becomes even more confusing when applying to AI systems since they are neither natural persons nor recognised as legal entities.
Since contract law is designed for human actors, it creates various challenges while dealing with commercial transactions which are delegated to autonomous AI systems.
AI’s challenge to Contract Law
As we have already established that the ICA is human centred, it is necessary to examine how the autonomous AI systems challenge these doctrines. The traditional contract law places significant importance on intention and consent. However, the autonomous AI systems may negotiate terms, even select contractual partners and carry out the contracts without any human intervention in a particular decision. Even if the system is installed by a human, the final contract choice is carried out by the AI system itself without any human intervention.
This creates uncertainty regarding the nature of consent. The ICA was developed with decisions to be carried out by the human mind, as intention can only be of a legal person. However, the reality of today with increasing autonomous AI systems weakens this concept.
Another major difficulty is due to the concept of agency. Even if AI systems appear to be acting on the behalf of their users, traditional agency doctrine also requires that the agent must be capable of acting within the scope of authority granted by a principal. Even though AI systems perform agent-like functions they do not have a legal status and hence cannot owe duties in a way that a human agent can. Thus, when the system produces results which were not intended nor reasonably foreseeable by either of the parties at such time the existing agency doctrine offers no guidance on how to deal with such situations.
At the same time, AI systems cannot also be treated as a tool. Advanced AI systems can generate outcomes that are neither predictable nor the ones which were specified by the user. The existing agency doctrine doesn’t provide on how the responsibility should be allocated in such situations.
In a situation where a company such as Walmart which uses AI-powered procurement and supply chain management systems to manage their purchasing decisions and such a system places an order which results in financial loss and a contractual dispute, then in such a situation who should bear the responsibility?
If we hold the developer of the AI system liable it would be unjust as the developer has only built the system and has no direct involvement in the transaction. The developer might not have foreseen every outcome generated by the complex AI system. Thus, holding developers liable would discourage future innovation.
If the AI system is to be held liable it raises the question of impracticability as the AI is not a legal person and lacks assets and the capacity to bear legal obligations.
The most efficient approach would be to hold the person assigning the AI system as liable. It is the businesses that choose to use the AI system, they benefit from using them and they exercise control over their operation.
The existing contract law doesn’t provide any clear framework for determining who should be held responsible in such a situation. This results in an accountability gap. Even though losses occur in the real world there is not a clear framework which identifies who is to be held responsible for the losses. This shows the limitations present while applying the traditional contract law to autonomous technologies.
New framework for AI contracting
The challenges created by autonomous AI systems cannot be solved by recognising AI as a legal person. On the other hand, it would further lead to significant difficulties. Instead, legal reform should focus on improving the mechanisms for accountability.
Courts can adopt the presumption that the contractual acts which are performed by the AI systems can be attributed to the humans that assign the work to them as they are the ones deriving benefit from using the AI system. By adopting such an approach there would be commercial certainty and at the same time it would ensure there are remedies for the victims of contractual breaches.
Thus, there is a need for clearer rules regarding the allocation of liability in cases of AI-driven transactions and to clarify the extent to which responsibility is to be shared among the developers, deployers and users.
Conclusion
Thus, the rise of autonomous AI systems in contracts has made it important that the ICA must also evolve to make sure that liability is placed justly and there are remedies for the affected parties. Since AI is increasingly being used in contractual transactions, the essential features such as consent, agency and liability are more difficult to satisfy. The existing contract law lacks a clear framework which can assign the liability for the acts and the breaches arising due to AI systems. As a result, judicial reinterpretation and legislative reform are necessary so that there is accountability and legal certainty. The ICA must also adapt to such a world in which the contractual decisions are no longer made only by humans.
About the Author
Dhuha Mir is a third year B.A. LL.B student studying at OP Jindal Global University and she is interested in women’s rights, contract law especially sports and international law.
Image Link: https://contractbook.com/blog/what-is-the-contract-ai-top-benefits-of-ai-contract-software