Posted By: Joseph Urtuzuastegui
Artificial intelligence (“AI”) has been the subject of many science fiction stories for the past fifty-years, but just how close is society to blurring the lines between who is a real person and who is a computer? We would like to think that it was still a distant idea but it is actually already here. AI has moved front and center into the business marketing world. There are many ways a business will use AI in order to market to target customers, and AI has become a great resource to many businesses. But, what happens when AI gets so good that the people who are talking to it think they are speaking with a real person? Furthermore, how do we handle AI entering into a contract with a real person who thinks they are talking to a real warm-blooded human? These questions will have to be answered in the near future, and the way a court will have to answer them is through the good old-fashioned common law of contracts. This analysis will determine how and why AI will be able to enter into legally-enforceable contracts with humans, but don’t be so quick to give up as we will discuss some defenses for us measly mortals.
In order to determine the legality of a contract entered into between a human and a “machine” we must first look to the elements of what makes a binding contract. Generally, for a contract to become legally binding it must have (1) an offer; (2) an acceptance; (3) consideration; (4) and capacity. An offer is exactly what you would think it is, promising to do something in exchange for something else. For example, I will give you $1 million if you sell me your house. The flip side to the offer is that there is an acceptance, which is easily illustrated by the above example if the homeowner says, “I accept your offer of $1 million for my house.” The consideration is usually just money or performance that is exchanged in order to give the contract a purpose. Again, using the example above consideration would be the $1 million from the buyer and the house is the consideration given by the seller. So far so good?
This next element of capacity is the element that needs to be focused on when there is a contract between a “machine” and a person. The question will arise, whether something that is not a living human being can enter into a contract with a real human being, especially when the flesh-bodied human thinks they are actually speaking with another person. Capacity, most simply put, is the ability of a party to understand that they are entering into a legally-binding agreement. Most times examples of incapacity include a minor or a person who is not of the right state-of-mind to enter into a binding contract. But, these are not the only two things that may lack capacity. For example, animals will lack the capacity to enter into a contract. So for anybody out there who wants to marry their pet, they will be hard-pressed to find an animal that understands what is happening enough to assent to the contract. But what about the AI? Does the AI understand that it is entering into a legally-binding contract? The short answer is, NO! AI does not think, it processes data and “learns” through data loads and the more data the AI goes through the “smarter” it becomes. It is a little (lot) more technical than that, but this is the basic idea of how AI works.
Ok, now that we know the basic concepts of how to form a legally-binding contract the first thing that should be going through your mind is, “well, if capacity is an element and AI cannot think how can they enter into a contract with one of us who have the ability of empathy and morals?” And you would be right to think that, but the reason the contract will be binding is a lot simpler than you might think. AI is primarily being used to make calls to potential customers for a business entity. Through these phone calls many people do not even know they are talking to an AI, in fact many people like talking to the AI rather than a real person. The fact that the AI is making the calls “for a business entity” is a very important aspect of the validity of contracts being entered. A business entity can enter into legally binding contracts with actual people through a legal concept which identifies the entity as a “juridical person.” An easy example to illustrate this would be if you are to drive to the grocery store and pick up a bag of chips, when you go the register you offer to buy the chips, the grocery store accepts your offer as a juridical person, and you pay consideration. There is no question of capacity in that situation, it is a common transaction that happens every day, you are not contracting with the cashier or the cash register which are extensions of the store, but rather, you are contracting directly with the grocery store. The concept is the same with entering into contracts with AI, you are not contracting with the AI bot itself but rather you are contracting with the “juridical person” who is under control of the AI.
Before we get into the defenses, I do want to point to one very important point. When we are talking about legally binding contracts that an AI presents to a person we are assuming that the AI is owned and operated by a business entity. If, like many sci-fi enthusiast believe, AI becomes rogue and independent the question of capacity would be a much closer call. But, at this time we will hope AI continues to be reined in by business entities that use them properly.
Looking to defenses of a contract entered into with the AI bot, there may be some defenses that may be an option. First we should look at the defense of unconscionability. If the bots are targeting a certain demographic, say older in age who may be more willing to contract with a stranger on the phone, there may be an issue of conscionability with regard to the contract. But, this is completely dependent on the individual who entered into the contract. This is a completely circumstantial defense, but this is important to be touched on to make sure the elderly do not get targeted by the AI bots. The difficulty to this defense will be proving that the individual entering into the contract would not entered into the contract but for the “person” speaking with them. The party must be wrongly induced into entering the contract. There may be the argument that had the person entering into the contract understood they were not talking to a real person, but rather a “computer” they would have never entered into the contract. The AI will most likely have a name that they go by, and the misrepresentation of them being an actual human may make all the difference in somebody entering into the contract. This may be a very hard sell to the court because many times the court will validate a contract if parties have agreed, but in the spirit of equity and fairness a court may grant some sort of relief.
The next defense, and I believe the stronger one, would be the defense of fraud. Fraudis present when one party materially misrepresents an essential part of the contract. This defense will be most useful if the AI never informs the contracting party that they are representatives of a certain business entity. If the contracting party believes that they are fully contracting with a person and not AI, and the AI never discloses who they are representing or a part of the contract that is very important the contracting party may be able to get out of the contract due to the material misrepresentation of the essential term. This defense will be very hard to prove unless there was actual fraud by the AI, which would actually be the company itself, it seems that as long as the AI discloses who they represent and discloses all the material terms of the contract there will be a valid binding contract.
To tie it all back together, it will be very important in the future for people to understand when they are entering into a binding contract because you never know who you are entering into the contract with. The Artificial Intelligence may be taking over, but we warm-bodies will be ready for them as long as we understand our contracting principles. If the AI is an extension of a business entity, they will be able to contract on behalf of the business, as long as the AI discloses they are representing the business (remember fraud). And, last but not least, to all the AI bots and the humans out there reading this when entering into the legally-binding contract, make sure you know who you are entering into a contract with and what you have agreed upon.