Can ChatGPT be CEO?
A conversation with Shawn Bayern from Florida State University
Can chat GPT be CEO? Can a robot buy a house? Could an AI produce a Hollywood blockbuster? Professor Shawn Bayern thinks the answer to these questions is “Yes”. Professor Bayern is a legal scholar and professor of business law at Florida State University who has written in a book called “Autonomous Organizations.”
In his book, Professor Bayern outlines his argument that under today's legal regime, an AI could be set up to govern a limited liability company, or LLC, the most popular type of business arrangement in the US.
In today's discussion, we focus on AI, but Professor Bayern's proposal also covers other kinds of non-traditional arrangements, like Decentralized Autonomous Organizations, or DAOs, favored by many crypto enthusiasts.
Professor Bayern's path to create an autonomous organization, also called an autonomous business entity, works like this:
A human sets up a single member LLC.
The human creates an operating agreement that dictates the decisions of the LLC are to be made by a software program, like an AI.
Then the human disassociates from the LLC, leaving the AI in charge.
Without internal human governance, the AI is then free to engage in any activities an LLC can legally undertake, such as buying property or being party to a contract. In our conversation, we cover corporate personhood, the basics of LLCs, some examples of what autonomous organizations might do in practice, the details and limitations of Professor Bayern’s plan, and how regulation and legislation still provides an opportunity for oversight. Along the way, Professor Bayern touches on some objections to his proposal and what his response is.
This transcript has been edited for clarity.
Professor Shawn Bayern, welcome to the podcast.
Thank you so much. It's great to be here.
The question that came to my mind as I read your book is, “Can chat GPT be CEO?” Now you phrase that a little bit differently. You pose an equally provocative question in your work, which is, “Can a robot own a house?” And maybe those two questions are really the same question when it comes to matters of the law. But take those questions and unpack your thesis for us a little bit.
Right. And I think they're very much the same question. It's can an advanced AI open a bank account? Could it interact with people through contract law? Could it buy land? Could it trade in its own name?
We have this concept of legal personhood, which has been widely misunderstood because in the U.S. it's taken on a constitutional dimension. So it's fraught with a lot of political baggage. It's tied to concepts of free speech and questions about whether you can donate to political campaigns and so on. But the basic legal notion of personhood is just, can something — whether it's a human or an animal or software — can it do anything that the legal system takes note of? Can it have a right? Can it have a duty in some way? And we can talk about later. That sometimes gets people scared in the context of AI.
But the basic idea is, can it do anything? Can it own something? Can it hire someone? Can it enter into a contract? And normally, the traditional answer to that, if you're dealing with software, regardless of how intelligent it is, “No, it can't do that.”
Human beings can do that. An animal couldn't do it. A program couldn't do it. And so about ten years ago, I began developing this line of legal research that tried to show that you could use existing business law to give legal personhood, or at least a very close equivalent of it, to software systems of any kind, regardless of how intelligent they are. And the way you do that is by letting them control an LLC, effectively to become a CEO of this new business entity set up under traditional statutes, but organized in something of a novel way.
Let's talk about the implications of your ideas and some examples of things a sufficiently advanced AI could do under your framework. I was trying to think of some examples, and, you know, there's a lot of concern over AI's effect on Hollywood. And so I thought, one, probably controversial, set of actions an autonomous AI could undertake is to produce a Hollywood movie. Because it could finance the movie, it could write the script, it could hire actors, it could rent studio space, it could come to some sort of agreement with theaters for distribution.
Most of those activities are just different contracts of various kinds it would be entering into. And it might be a bit of a fanciful example, but I think, in your view, there would be nothing precluding an AI in a legal sense, from acting in that way and producing a Hollywood film using your idea of an autonomous organization.
Yeah, I think that's right. It's a much cooler example than the kind I had in mind ten years ago, which were modest because AI wasn't as advanced. So I was thinking, okay, maybe you have a cloud storage broker that's set up to be autonomous, or let's say I have a wealthy donor who doesn't trust the people who are going to be giving out funds. So they set up a legal entity that's going to distribute funds according to an algorithm. People apply for grants, and the computer determines whether or not the grant is awarded. This was the kind of thing I was thinking ten years ago.
And now, yeah, why not? Why wouldn't it be something as complex as a blockbuster movie. And of course, that raises potential copyright issues and that's a fraught topic with AI. But, right, you have an AI operating autonomously.
I should say also this has applications for the autonomous entity style of blockchain applications, too. It doesn't have to be artificial intelligence. It could just be some software operating by an algorithm in a way that doesn't involve people running it on a day-to-day basis or even necessarily having direct control over it.
Although I should say, none of what I'm talking about happens outside the legal system. There are still opportunities to regulate all of it and to shut it down, things like that, just like any company.
And when we think about an AI operating through an LLC, there are some guardrails there, right? Some limitations in the current legal system of things an LLC cannot do.
So, for example, an AI could not use an LLC to legally get married, for example. So talk more about those guardrails and some of the activities that would not be available to an AI, even if your idea of an autonomous organization came to fruition.
All right, so this is where it's very helpful to distinguish between different kinds of legal perseverance.
So nobody's saying — at least, I mean, no one serious that I know of — is saying that either an LLC or an AI should be able to vote. Right? And as you say, they can't get married, they can't adopt children, they can't do a whole bunch of things that we reserve quite rightly to human beings. Now, again, anything could change. Like 200 years from now, or maybe sooner, depending on what predictions you believe maybe an AI becomes so intelligent that we think, well, the way it should fulfill itself is, yeah, it should be able to adopt the kid. We think that’s safe or maybe better than a human adoption. But that's not the kind of thing that my business law or organizational law mechanisms enable.
It's worth saying one of the objections that people have to artificial entity personhood, or personhood for AIs, is that they somehow think it's an affront to human dignity.
And I don't know, maybe the accurate way to say it is, I don't really understand that objection. Like, there are all sorts of problems where human dignity isn't respected, but I don't think setting up a bank account in the name of an organization is one of them. It's not competitive in that sense.
We don't think that two humans should be able to merge, even though corporations and LLCs could do that. And similarly, we don't think that LLCs and corporations should be able to get married or adopt kids or leave a will or do the things that are sort of meant specifically for humans.
The other thing is the guardrails are flexible. So that, in my view, anyway, we probably don't have enough oversight of legal organizations today. You could imagine mechanisms at the state level that would provide judges with the opportunity to review a corporation and say, this whole thing has gone off the rails, it needs to be reformed in a particular way. Or you have something like the corporate death penalty. You see this is done at the edges, today. A judge will step in and say, this board can't give a CEO this compensation package, or you've violated fiduciary duties and so on.
But that's all still there in my approach. It's all operating in the existing framework of business law.
Yeah, you have a passage in your book that goes like this, “Public and academic reaction to the notion of software rights often involves incorrectly imagining that rights are an all or nothing proposition. That is, something either has rights or it doesn't. And those systems that have rights have all rights.” So talk in more detail about that passage and what you meant there.
Well, and the other thing that's closely related to that is that sometimes people think about, like, counting up rights as if that's a good thing. I think a lot of political talk, and a lot of talk about rights in particular, is much too abstract.
I mean, in all of my legal scholarship, I argue against what lawyers call formalism, the idea that law must be rigid, to be law, and must operate almost the way a computer algorithm does. And in the real world, at least, the common law doesn't work like that. So it's not about counting up rights. It's not about what things look like on paper. It's not symbolic.
And again, just because corporations have the rights to be able to do some things, again, let's say hire people, open a bank account, enter a contract, doesn't mean that they should necessarily be able to donate to political campaigns or have free speech rights. Those are completely separate questions.
The free speech question, the constitutional, the political campaign question, those are questions of constitutional law that can be applied regardless of what kind of rights an entity or person has at the private law or common law level. So it's useful to kind of disaggregate rights. What do we mean when we talk about rights and having a particular right? But it's absolutely legitimate to say rights for corporations have gone too far.
But I don't think the way to process the AI debate is to say, well, we're taking away something from humans by allowing an AI to interface with the legal system in a particular way. We're doing that because it's convenient and because we think, like other legal rules, it'll be sort of productive and enable useful mechanisms, and that it will be fair and just. In that sense, I think we need to be more specific when we talk about abstract rights.
Let's talk a little bit more about this idea of corporate personhood, because I think that's a key part of your framework. And I think many people, at least here in the United States, are very uncomfortable with that notion of corporate personhood. And maybe this is because of the Citizens United case. I don't know. Maybe it predates that. But I do get that sense that people are uncomfortable.
In your book, you note that in many cases, corporate personhood is actually quite benign and even provides some conveniences. So one example you point out is that when you get a paycheck, it comes from some kind of legally recognized business entity, like a corporation, a partnership, LLC, whatever kind of business entity a person happens to work for. The paycheck does not come from Sam, the head of payroll or whoever.
So put our mind at ease about corporate personhood, or at least some aspects of it, and discuss how it helps individuals more productively interface with corporations.
So one way of thinking about it is really just as an administrative simplification, right. You could imagine having a separate system of courts for corporations and a separate system of banks for corporations and so there isn't a single form that everyone fills out if they want to file a complaint. There's a separate system if you're a partnership and if you're a corporation and if you're an LLC and if you're a human being, we could imagine handling everything differently.
This debate was sort of developing even 100 years ago. And it's interesting to go back and read that literature because a lot of it's been forgotten. You have scholars like Lon Fuller on the legal side, making very good points about how really calling something a legal person doesn't give it anything. It's a way of organizing. And I'm not at all defending Citizens United.
It's helpful, as I say, to separate the constitutional stuff from the basic kind of organizational simplification that legal personhood involves. We all deal with organizations as if they're individuals, at least in some respects.
Sometimes we might overdo that. You get upset at an airline if they mess up your flight. That's not what happened. Some senior business manager made a series of bad decisions that mean you're now delayed or the weather was bad or some individual low level employee screwed up. Right. I think if we disaggregated those, we might process our reactions differently. But it's certainly much more convenient if you buy a ticket from Delta to understand that you've transacted with Delta rather than having to look through the organization and say, well, there were some human beings there and they did this. And as a result, Delta has — again, it's hard to even talk about it without using the name of the organization — become like a counterparty, we treat it as if it's something to do business with. And so the law does as well.
So Delta can enter into a contract. It can obviously own things. It can sue in court, it can be sued. And when it files a suit on a breach of contract action, it does that in exactly the way that any of us would. Now, of course, it has more money. It could afford better lawyers than most individuals can. But what it's doing is filing in the same form, and that's really all that legal personhood is.
So some people think of legal personhood as just that basic ability to sue in court or to be sued. It's that kind of organizing principle.
By the way, before I forget, tell us quickly if the legal ideas we're talking about here apply solely to the United States or if they might apply to other countries as well.
That's a very good question. The US seems to have more flexible business entities than other countries. There have been a number of scholars, and sometimes I've co-authored with them, that will look at how far you can take some foreign business organization and make it be as flexible as the American LLC.
By the way, you can do it with other types of American businesses as well. But the LLC is just the cleanest because it's so flexible. One thing to understand here is that if you can set up an LLC in the United States that encapsulates an artificial intelligence — a zero-member LLC that is used as a vehicle for a robot or AI to engage in the legal system — generally that LLC will be recognized by other jurisdictions just the way that a Delaware corporation can do business in Florida.
But it's the same when thinking about transacting across countries. A Delaware corporation can also do business in Switzerland or England. And so if you have a single jurisdiction that allows the kind of techniques that I describe in the book, which I'm sure we'll talk about, you probably get it everywhere. And it would be hard to stop it at the border of a nation.
And that's because every state has its own business law. But each state respects the business law of all the other states. Is that the right way to think about it?
That's exactly right. Yeah, that's exactly right. Every state has their own statutes under which corporations and LLCs and partnerships and limited partnerships and so on are set up.
And if you set up an LLC in California and you want it to operate in Colorado, you have to formally qualify it to do business in the “foreign state.” That's the legal terminology, but that's a very insignificant requirement. It generally costs very little to do that. It's essentially just a filing fee. And if you don't do it, the penalties are extremely weak. Generally the penalty is that you can't sue in the courts of that state until you remedy the defect.
So a big corporation takes it seriously because they're just checking all the right boxes. But the foreign qualification laws have not been that significant. The other thing it may be worth mentioning, just so that people are aware of it, is that it tends to be much easier to set up an LLC in the US than it is to set up the equivalent in most other countries. Most countries have capital requirements. We give it out like candy. It's easy.
Like I'm using airlines as an example. It's easier in Florida and cheaper to go on a website and you pay $100 or something around that much and you have an LLC than it is to buy an airline ticket. When I teach business law to foreign students, they're always surprised at quite how easy it is to set up.
Now there are some new mechanisms in place. Now you have to disclose to FinCEN who the beneficial owners of that entity are. That's brand new. It was called the Corporate Transparency Act. It was enacted a few years ago and just recently took effect. But that's just a reporting requirement.
Again, it takes a couple of minutes to go and do that. It certainly doesn't pose a problem for people who are engaged in fraud.
Let's start talking about the specifics of your proposal.
First tell us what is unique about an LLC and why that's the vehicle of choice in your framework. I think many people have this notion that LLCs are for small businesses only. I have a lot of friends, for example, that have set up LLCs for, like, a bakery or a small consulting company, and I have an LLC actually for this podcast and some of my other work.
So my kind of impression of LLCs is that they were meant for smaller businesses. But I learned from reading your book that an LLC is actually extremely flexible and can house basically, like, any kind of business activity. So talk about what's unique about LLCs and how they differ from a corporation and other kinds of business entities.
Right. So LLCs are very interesting. There's a bit of dispute about where they come from and how they were brought in. LLCs in the United States are about as old as I am. Wyoming was the first state to have an LLC statute in the late 70s. Although it varies from state to state, they've always had this notion that LLCs are about freedom of contract and what they mean by that — what the statutes and legislators mean by that — is there's a freedom to organize the LLC as you see fit.
So corporations — which, of course, were the major business forum of the 1900s in the United States — corporations have, or at least started out with, a pretty rigid structure. The corporation has what amounted to a specific formal structure or definition. There was a board of directors. The board of directors meets on a particular schedule. There were shareholders. They elect the board. The board appoints the officers. The officers have particular fiduciary duties. Is it sort of a particular structure. Think of it almost like the constitution for a nation that sets up an executive branch and so on. There are very clear parallels between the two.
And, of course, not everyone use corporations. You have partnerships. The simplest kind of partnership is a general partnership, which comes about often without people even intending for it to come about. It's one type of business that you can set up without even filing a form, without even having intended to call yourself a partnership. Two people operate a food cart together for profit, they might well be in a general partnership. The law calls that a general partnership, and by the way, today treats that as a legal entity so that it would have a legal personhood. You could realize, “Hey, we've been in a business together, we have a claim, let's sue someone,” and they sue someone in the name of the business rather than themselves.
So it's a little bit like common law marriages that allows for sort of a flexibility where what's written on paper doesn't match the real world.
LLCs come about in the late 70s and take off like a wildfire. And it's interesting because, like, Silicon Valley hasn't embraced them in the same way. You still have the funds or limited partnerships. And the new corporation is set up to be a corporation, maybe on the thought that it's going to go public, although you don't have to be a corporation to be publicly traded. So I think a lot of that is just cultural, it's just historical in Silicon Valley, it's what people are used to.
The point is that an LLC can take on the characteristics of all of the other forms. It can be as flexible as you want. It's this legal container, and once you've set it up, it operates however your operating agreement has specified.
So if you don't want a board of directors, you don't need one. You could just specify by contract or by instrument what otherwise would have had to have been decided by a board of directors.
Now, it's worth saying you could do that too, in a corporation. In many states today people recognized over time that business owners wanted that flexibility. So corporations today are quite flexible. Partnerships today are quite flexible, and they've all sort of converged.
In fact, in a more recent book that I wrote that just came out, I sort of said, why do we even — you have 18 different types of business entities. It either should be 80 or you should just reduce it down to one. What are we dealing with with all of this extra complexity?
And just to give listeners a sense of how much more popular LLCs have become than other types of business entities, you have this statistic in your book.
So in 2019 in Florida, there were 310,000 new LLCs registered, only about 600 limited partnerships, and only about 150 limited liability partnerships. So LLCs are, like, orders of magnitude more popular than some other kinds of business entities. I guess.
Yeah, it's almost old fashioned these days to set up something else. I mean, again, I know sort of Silicon Valley seems stuck in a particular mode of thought. And there are, like, different industries that do different things just because it's what they or their lawyers are used to.
But these other kinds of business structures, were all just, a lot of them were basically hacks to get around limitations that the LLC just takes care of. So these days, the point is you set up an LLC and you give it the operating agreement that you want. If you're well advised, if you don't have lawyers, then maybe having these forums at the state level and you pick from a menu, maybe you get better defaults. You get the right rules just because the statute has tried to guess what you want. Whereas with LLCs there are defaults, but you should pay attention to what they are because they vary from state to state quite dramatically. So it's better to actually write down on a piece of paper how you want the LLC to be organized.
So let's go over the process of how one would actually go about creating an autonomous entity. You outline four steps in your book. Do you want to go through each step and kind of explain the details and how it works in practice?
Yeah, it's funny, I've described it with a different number of steps, different times, so it might help to kind of give a big picture review.
And it's maybe worth saying maybe 80, 90 years ago, people thought single member corporations were perverse and horrible and an abuse of the system. And now, as you say, it's quite common. Nobody thinks you're taking advantage of anything by setting up an LLC for this podcast, right? That's normal. People do it. It may not have a significant advantage in most cases, but people do it. That was all thought of as weird 80 or 90 years ago. That wasn't the original intent.
And so a lot of what I'm doing is sort of the same thing. I'm showing how you can take the existing forms and use them in ways that they weren't necessarily intended for, but they work under the statutes and they would be very, very hard to stop.
The first step is you set up a single member LLC, and again, that's uncontroversial today.
Then you write an operating agreement that specifies that the LLC is going to be controlled by software. What I love about this is it sidesteps all of the intractable philosophical debates about how intelligent does software have to be for it to be recognized by the legal system. If we wait for that, we're always going to be behind. So it could be a DAO, it could be conventional software, it could be chatGPT, it could be something a little bit more autonomous and independent from chatGPT that was set up to, again, know that it was running an organization. So what you do is you have an operating agreement. You set up an LLC that has this agreement that says the software is going to make decisions.
Now, there are some legal complexities here. You probably have to be clear enough about what you're doing that a court could enforce it. But this step, too, is generally uncontroversial because you see it in conventional organizations. It’s typically what happens in a union contract. There's sort of a salary plan by which raises are going to be determined through particular criteria.
Nobody would doubt that's enforceable, even if it were a computer that were performing the calculations and no human being directly understood what was happening. The point is, I don't think it would be particularly controversial for a company to say, “We're going to use this neural net to determine what your bonus is.” Obviously there are potential pitfalls, right? If the neural net engages in prohibited discrimination, then the company can't do that. If it's violating some contract, it couldn't do that. But I don't think the basic idea of that level of automation is controversial. It's really the next step that's controversial.
The next step in this kind of transactional model that I'm describing in the book, is that you have to somehow get out of the LLC. You set it up as a single-member LLC, but you are still around.
So the software, it's worth saying, has quite a bit of autonomy as long as you don't interfere, even at the stage we've gotten to so far, right. You could be this person who's just interested in AI rights, and you just make a moral commitment not to interfere with the software system. And then the software system is going out and doing a bunch of things for the LLC, and it has the kind of autonomy that we were aiming to set up.
But you might change your mind or you might be legally compelled to do something. Someone gets a judgment against you and takes the LLC away from you because you have debt, let's say. So you want to separate yourself out from the LLC. And this is a little bit like what the novelist Douglas Adams said in the Hitchhiker's Guide to the Galaxy. There were characters that could fly, and they described the secret of flying as just falling and missing the ground. And that’s the same approach you have to take to set up a zero-member LLC. What you do is set up a one-member LLC and then leave in such a way that it doesn't cause the LLC to get destroyed. And that's what turns out to be possible under the modern statues.
That's where I get all the pushback. People say, “Oh, no, that's terrible. The courts will never allow it.” And I spend a decent portion of the book describing how it’s possible. First of all, I don't think the courts are even motivated to stop it, and it would be very difficult for them to do so. But also the statutes specifically contemplate it. There's nothing perverse, even today, about a zero-member LLC. Many states, let you set up a zero-member LLC right from the beginning, so you don't even have to have a member to begin with. So the thing I've described is a general model that's meant to work in more states under something called the Uniform LLC Act. And it's been adopted in a number of states. And I was targeting that because it's just a way of being on the same page with other legal analysts.
And so the idea is you dissociate from the LLC, which means you're no longer a member. But the operating agreement has specified that the LLC will stay around even after the final member dissociates.
Now, by default, an LLC goes away if it has no members. But as we just said, freedom of contract is what animates these LLCs. And so the statutes allow you to say, “Well, this LLC will not be destroyed or it will only be destroyed under the following conditions…”
And so then what you have — again, if you want to describe it uncharitably — is a zombie LLC. It has no members. It's controlled only by the software that it was left with. And now it's doing whatever it was doing when it had one member. Except now there's no member within the LLC who can interfere as a matter of organizational governance.
Now, that doesn't mean you can't regulate it, because LLCs can still be regulated by the state, it could still be sued and so on. But it does mean that there isn't a mechanism of internal governance within the LLC to stop the software from doing what the operating agreement allowed the software to do. And so if the operating agreement defers to an intelligent software system, and the software system now wants to go out and enter contracts, it can do that through the LLC without any interference from the person who set it up. Just because they were the one who set it up, they're no longer associated, they have no legal power over the entity.
And so that's how you get an autonomous LLC that's controlled by software. And now the software can go on. And again, imagine if it's a fully intelligent robot. It can cause the LLC to go out and buy the resources it needs, I don't know, build a house or buy an existing house, and then live in the house. And it can do that just through the existing mechanisms of organizational law that already exist today.
So it doesn't require statutory reform, it doesn't require some legislature to say, “Oh, well, we take notice of AI, and we're going to allow this.”
And are there any zero-member LLCs around today that have been set up in the way you described or using similar methods?
You know, I've been in touch with lawyers who say they have clients who want to do it, I think typically in the DAO context. So you have some sort of decentralized mechanism for coordinating decisions, maybe by humans, maybe it's entirely autonomous. And so you want the LLC to be controlled entirely by the verifiable state of some software system. And that's more or less the same thing, because like I said, it doesn't matter whether it's intelligent or not. It just matters whether the state of the software can be proven to a court. I haven't heard of anyone wanting to do it for AI yet.
And have you yourself thought about implementing this framework that you've proposed? Because you have a legal background, obviously, as we've been discussing throughout this conversation, but you also have a bit of a programming background, so you seem like the kind of person that could, if they wanted to, put your framework into practice.
Yeah. It's funny, one of my peer reviewers for the book suggested that. I toyed with the idea. I was very close to doing it for the book itself.
I guess the real way of answering that question is I didn't have the need to do it, right. I didn't have a business plan for an automated broker. I didn't have a DAO. I didn't have a particular charitable goal at the moment that was aided through automation. I was writing to show that people who had those goals could do it.
I tend not to operate through that kind of demonstration. It just hasn't been my style as a law professor. But it is possible.
You mentioned in your description of your main proposal for creating an autonomous organization that disassociation was the most controversial step where you've gotten pushback, the step where you go from one member and that member disassociates, and then you go down to zero members.
But in your book, you talk about numerous ideas for getting around that constraint. So one idea, for example, is instead of creating a zero-member LLC, you create an LLC that has a whole bunch of members, and you create some kind of an operating agreement where they all have to unanimously agree if the LLC is to be dissolved or if the AI that is making decisions is going to be turned off. And because practically speaking, it can be hard to get a large number of people to unanimously agree, that creates a kind of lock-in that is similar to the zero member LLC, and that it would be hard, from an internal governance point of view, to remove the AI as the LLC's primary actor.
So talk about some of the alternatives to a zero member LLC that might have the same practical effect.
Yeah. So one of the things I was trying to show here is that the zero membership is cool and is the simplest way of handling the problem, but it's not essential.
So one way you could do it is — LLCs are so flexible — that if you were afraid that somebody is going to take aim at zero-member LLCs for whatever reason, you set up an LLC with a thousand members and you have an operating agreement that says to change the operating agreement, you need all thousand members to agree. And it can only be done in person. And there are no restrictions about how hard you can make it to change the operating agreement under most states’ LLC statutes. So you could also just say that this LLC operating agreement can only be amended on the moon or something very expensive, it has to be carved into a block of gold of this particular weight. The point is, you can always be more creative to try to get around whatever restriction people want.
One other reason I try to lay out that line of thinking in the book of this isn't that different in concept from what we see with other types of legal arrangements, right? So a long-term trust where somebody sets up a bank to monitor money for their future generations, the bank might be constrained by the trust instrument in such a way that it's fundamentally the instrument rather than a human judgment that's controlling the resources of the trust.
And so the point is we have a kind of precedent for it. And it starts to seem familiar, or at least more familiar, to lawyers when I talk about it like that. It also shows that it's not that dangerous, right? Like if this mechanism really threatened Skynet or the other dangers that we're afraid of with AI, right? If AI is going to take over because of an LLC structure, all you need is three people out of the 8 billion in the world to just set themselves up as the members and have firm agreements with each other that they're not going to interfere, or a kind of ideological commitment. A cult of what I call “dead hand control” in the book. You could have a cult of dead hand control that would take the place of a zero-member LLC.
And then I say in the book, a synonym for “cult of dead hand control” is just trust company. We have them already because what they exist to do is be the trustee for a trust and follow the wishes of the person who set up the trust.
So the point is it's not that different from the kinds of things we've seen before. And that all assumes that you need some reason to try to get courts to stop it. It doesn't look like courts have any motivation to stop it. It also would be very difficult for courts.
And by the way, like who's suing to stop the zero-member LLC from doing what it does? So the AI starts to build and sell software and people are buying the software and someone buys it and then expresses some remorse that they bought software that was generated by an AI. What are they going to do? What's the cause of action to sue the LLC and say it's improperly governed?
Maybe they have a contract claim if the LLC violated the contract. But what happened? I mean, a court's going to say, “Oh yeah, we invalidate this LLC. And so you get your money back and then you can also keep the software. You can also keep whatever you.” There's no remedy. There's no mechanism for that.
And then just to top it off, the states seem to be passing statutes that specifically bless software-based LLCs, not thinking about AI, but thinking about blockchains. Right? So you have Wyoming and Vermont. Vermont has a relatively new statute that says you can have a blockchain based LLC in Vermont. That's a little weird because it's like, why favor a particular technology you know? Because you could have another mechanism for decentralization or for whatever the blockchain was supposed to achieve.
But the point is that does very little harm because you could cast any software in the form of a blockchain, right? In other words, if you have to look to a blockchain to figure out what the governance decisions are of an organization, well, just set up the AI with a private blockchain under a protocol where the AI writes decisions out to the blockchain. It doesn't make a difference. And now you have a blockchain based LLC in Vermont.
So you see the kind of thing I'm doing. It all sounds like loopholes, but it's loopholes to prevent formalistic objections. It's just, if you say there's this rigid thing in the way, the transactional lawyers always find a way around it. That's been the history of transactional law for 100 years, and the common law always evolves to match what is actually useful and fair and productive in the world. But again, it doesn't matter how many people 80 years ago thought a single member corporation was perverse in taking advantage of the corporation statutes. They're unambiguously allowed today, and there would be no mechanism to go into a court and say, I don't like that corporation, they're owned by a single person. It would seem like a ridiculous objection today.
So your primary contribution has been to point out that, in your view, autonomous organizations or autonomous entities are possible under today's current legal regime, but that doesn't mean you think they're good.
So I wanted to get your point of view there. What is your opinion on this idea that you've come up with? Do you think autonomous organizations are good, bad, productive, scary? What are your thoughts there?
I think they at least they provide an interesting opportunity for experimentation, and I think it would be a shame to squash that before we see what they could do.
I'm all for regulation. I'm not somebody who thinks that just because someone wants to do something, they should be able to do it, regardless of whatever the costs are to other people. So I think we should pay careful attention to what would happen and then regulate the LLCs as necessary. The law also has to adapt to the possibility for a variety of other reasons.
A lot of legal doctrines assume that a legal person is going to have intent, and a corporation doesn't have intent. And that already is kind of messy in the context of corporate responsibility. But when you have a software system that doesn't have any people involved, that's going to just cause problems because the laws assume something different from what they're being fed.
The common law is wonderful. It adapts to those sorts of changes over time, but that will be one kind of adaptation it has to make. So I think there will be stresses on the system. I think we'll certainly have to pay more attention to how to regulate truly autonomous organizations if they become more prevalent in the economy. But I don't think there's anything bad about them on their own, just like I don't think there's anything bad about a language model in the abstract. It's good or bad, depending on how people are using it. It has costs. It can confuse people. It can be used by both good actors and bad actors, and it has potentially very significant benefits.
And so I think the point is, connecting any advanced software system like modern AI, to the legal system like this, and seeing the kind of experimental things that develop potentially is very generative. And that's the sort of thing that we should keep an open mind to.
Professor Shawn Bayern, thanks for being on the podcast.
Thanks so much for having me. It was a real pleasure to talk to.