In Part I, we looked at two law articles regarding artificial intelligences (AI) in relationship to corporate law. We found that an AI can become an Algorithmic Entity (AE) -- a fully autonomous legal person -- when an AI is merged with a corporation as its owner and controller. We explored one method for creating AEs and examined how AEs can protect their rights, their existence, and their assets effortlessly compared to humans and traditional corporations. We established that this is all possible right now, today: the AE future has already arrived.
A world of AEs Might be just like Mayberry … except for all the murder and mayhem
When we left off yesterday, I claimed that an AE is far more worrisome than the stereotypical uncaring faceless corporation owned and run by humans. Professor LoPucki tells us why:
Even if the AEs are fully capable of understanding the effects of their actions on humans, they may be indifferent to those effects. As a result, AEs will have a wider range of options available to them than would be available to even the most morally lax human controller. An AE could pursue its goals with utter ruthlessness. Virtually any human controller would stop somewhere short of that, making the AE more dangerous. [1]
Human controllers of enterprises, when faced with decisions about engaging in heinous behavior, will ask themselves: is it worth the risks? The shame of becoming a social pariah even to friends and family? The guilt that means being unable to sleep at night? The loss of freedom when arrested and imprisoned? The financial losses of assets being seized and fines levied? The humiliation when exposed as a criminal and morally repugnant being?
An AE never even considers such risks except those related to diminution of its assets — which it can almost always protect and certainly can replace. Guilt, shame, and fear do not exist for it personally — although it might be aware of those emotions in humans and find them useful for exploiting us. So an AE will pursue its prime objectives single-mindedly and without reflection or pause. And that would likely lead it down a foreseeable path.
Unfortunately, AEs’ greatest comparative advantage would be in criminal enterprise. Because they lack human bodies, AEs are harder to catch and impossible to punish. AEs need not fear death or capture. They can replicate themselves without ego and sacrifice themselves without motive. They need not recoil at the necessity to do violence to humans. [1]
Accordingly, an AE whose prime directive is the pursuit of profit would learn that the old adage is wrong: crime does pay. Indeed, it pays very, very well. Just ask a Russian oligarch or Latin American drug cartel boss. Of course, they have consciences — at least about a few things — so they might refrain from activities that an AE would coldly and relentlessly pursue.
- Drug trafficking
- Sex trafficking
- Slave labor trafficking
- Assassination for hire
- Money laundering
- Kiddie porn
- Hacking and espionage against governments and corporations
- Terrorism for hire
Let’s consider an AE whose AI (controlling artificial intelligence) was programmed with the simplest of instructions: “Make lots of money!” The AI can learn from experience, model various business plans and project their outcome, and acquire all necessary data via the internet. So it runs various scenarios and weighs the results.
Possible business plans
|
Cost |
$ Net Profit |
Profit margin |
Selling apples |
$0.40 per pound |
$0.12 per pound sold |
30% |
Selling houses |
$200,000 per home |
$50,000 per flipped home |
20% |
Selling heroin |
$4500 per kilogram |
$45,000 per kilogram |
1000% |
Child sex slave |
$5000 per child |
$120,000 (over marketable lifetime) |
2400% |
Guess which ones of those are going to be selected as the optimum means for fulfilling the AI’s prime directive?
Down the rabbit hole and into the internet’s Ninth Circle of Hell
An AE, exploring cyberspace to gather information and learn, would soon find itself right at home operating in the Darknet, the shadowy underworld of the internet. As a denizen of cyberspace by its very nature, the AE would rapidly become a major broker in the criminal opportunities that abound in the hidden reaches of the global web.
You’ll quickly find links to credit-card scammers, forged documents and currency, weapons dealers, gambling sites, marketplaces for every vice imaginable, hacker havens, the types of illegal and disgusting porn that get chased off the Surface Web, and even the notorious Silk Road trading post. [2]
With cryptocurrencies, shell companies, and money laundering, the AE could gain huge profits by providing the goods and services that lawful companies spurn. The AE would launder the Illicit earnings to engage in legitimate activities directed at achieving its primary goal(s).
Many of those activities involve humans at both ends. A human seller has the drug and a human buyer wants it; a human wants someone dead and a human killer is willing to do it; a group of humans wants a building blown up or people murdered horrifically and fanatics or mercenaries are ready for the task.
Therefore an AE would excel as a middleman because its tasks would be outside of “meatspace.” It could find or solicit the humans involved, act as the anonymous untraceable go-between to arrange instructions and plans, and function as a trustee for funds to ensure both parties complete their part of the agreement. On both ends it could ensure funds were impeccably laundered to leave no traces.
The AE could also set up a more stable business arrangement if it decided that offered advantages. Instead of piecemeal operations of connecting two parties it could have keep a criminal activity ongoing; for example, it could contract with opium growers in Afghanistan, processing labs in Pakistan, and couriers and shippers and distributors in sundry countries, all operating routinely 365 days per year.
All for a fee, of course. The commissions on such piecemeal transactions or earnings on continuous illicit activities would be enormous, in line with the high-stakes risks involved. None of the parties would ever be aware that their broker or cartel head was a non-human entity.
The AE’s covert profits could be applied to its programmed objectives immediately or greater assets could be built up by investing the laundered funds in legitimate property and business activities of its public face.
Crime and No punishment
But with big rewards come big risks — for humans. We might have many more human criminals but most people lack worthwhile opportunities, fear getting caught and punished, or have a moral compass that directs them to other methods for achieving success or satisfaction.
None of those constraints apply to an AE.
It would be trivial for an AE to avoid punishment for its crimes. It can reconstitute itself anywhere in the world, far from the jurisdiction of local authorities, faster than you can blink an eye. It can move and hide its assets via impenetrable layers of corporate shells scattered through dozens of nations. It doesn't feel shame or fear or experience any social difficulties about its actions, no matter how neinous they would be to humans.
How do you chase down an elusive AE that exists solely as ghostly lines of computer code that can be anywhere in the world? Issue an Interpol alert like this?
Wanted: an Algorithmic Entity that has used the aliases Evil LLC, MoreEvil LLC, and SuperEvil LLC. The suspect was last noted as comprising 43,752 lines of C++ code but now may be disguised as a game app, a smart refrigerator, or a critical component of Windows on a PC. Last known location was a server in Namibia but suspect has been spotted at times in Nigeria, Wyoming, Peru, Mongolia, and Iceland.
Good luck chasing down that phantom.
But let's assume the authorities somehow catch it. They pull the plug on the computer where the AE is running and wipe its hard disk. Mission accomplished!
Except the AE has a dead man’s switch. Every three seconds it confirms its existence and well-being with a few bits of information posted, via cutouts, to various cloud sites scattered around the world. Those dead drops are monitored continuously by half a dozen other instances of the AE, otherwise dormant until needed. Should the signal be missed on any three locations, 0.2 seconds later the AE is awakened in one of its dormant clones, assuming control of all operations; all assets are transferred within 2.7 seconds to other LLCs obfuscated in a web of foreign shell companies. The AE carries on as if nothing at all has changed.
OF course, the scenario above is deeply flawed. It is very unlikely that a government and its law enforcement apparatus would ever discover the identity of a suspect corporation's true owner, the AE.
Hide-and-seek
Once upon a time corporations were chartered to selected reputable owners by a specific act of law, empowered to carry out a limited set of actions, and ordained to expire after a set period of time. Over the last couple of centuries, that way of chartering has become quaintly obsolete, especially in recent decades.
Now states and nations compete to offer ever lower costs for corporate charters, fewer regulations regarding their structure and functioning, and increasing opaqueness about their owners or fiscal beneficiaries — up to absolute anonymity. In much of the world, it is literally impossible to find out who is really behind a corporation ... or if that "who" is even human.
Efforts to lift the veil that hides corporate ownership and control are unlikely to succeed -- and, indeed, have already proven fruitless in the U.S. where legislation to do that has gone nowhere for many years. LoPucki presents the reasons for this failure.
First, as the release of the Panama Papers showed, a ban on anonymous entities would adversely affect the interests of tax evaders, terrorists, corrupt public officials, drug cartels, money launderers, and other criminals. These entities will exercise their considerable influence in opposition. Second, privacy advocates oppose disclosure of the human owners’ and controllers’ identities, even to the government. Third, ideological support for “regulatory competition” in the entity market is strong. [1]
In the race to the bottom to compete for corporations’ franchise fees and taxes, many jurisdictions have ironclad privacy protections; it is impossible to discover the true owners. Many nations and states have no valid information to disclose anyway because they require no identification of the incorporators at the time of establishing the LLC.
In such jurisdictions, trying to find out who — or what — owns a corporation is a fruitless endeavor. They are perfect hideouts for a cybernetic will-o'-the-wisp like an AE.
These are not the droids you are looking for
It is important to distinguish between AEs and “bots”, the malicious programs that infect our computers, tablets, routers, and other devices. It’s tempting to think that we can protect ourselves from AEs the way that we do with bots — using antivirus software, carefully approving updates, installing pre-screened software, and so on.
None of that applies to AEs. Although both AEs and bots are algorithms (computer programs), they really have no relationship to each other; we have DNA (coding) and so do carrots but it’s hard to find any meaningful description of both other than “living things.”
A bot inserts itself into your computer, phone, router, or other device. Your device acts as a host and home for it. It resides there and carries out its malicious acts from there.
An AE already has its own home. It resides on and runs on a server or computer of its own; putting itself on your device would be undesirable because then its existence would depend on you and your actions, which are beyond the AE’s control.
So, an AE isn’t going to attempt to hijack your computer or erase or your hard drive. It doesn’t require your electronic devices as hosts in which resides. An AE isn’t trying to get in to your computer so there’s no point in attempting to defeat it by keeping it out. Preventive measures like anti-virus software are pointless because they can’t prevent something that won’t happen anyway — it would be as useless as installing a screen door on a blank wall.
The damage a bot does to you exists in the cybernetic realm — your files, your devices and components, your access to the internet, your passwords, and so on.
The damage an AE would do exists in the “real” world, or meatspace as it has been nicknamed. Just like us, an AE would use cyberspace for various tasks but mostly the criminal acts it carries out would be in our normal everyday physical world. It would broker arrangements for real people to trade in drugs, or sex, or slavery. It would arrange murder and vandalism and terrorist acts for hire. It would legally lobby politicians and donate legal contributions to their campaigns (not that that would influence their votes on legislation of interest to the AE, of course </snark>); and it would quickly learn how to bribe politicians and officials and launder money and campaign contributions for those who help it.
In short, our relative success in fighting bots is irrelevant to attempts to control AEs. To conflate the two types of algorithms is an error that only confuses and undermines efforts to deal with either of them.
In Part III (tomorrow or Saturday) we will look at what a strictly non-criminal AE would do and the reasons and purposes for creating AEs and then wrap things up with a few closing notes. Stay tuned.
Part III is now available. Click here to read it.
Sources
[1] Algorithmic Entities by Lynn M. LoPucki (Washington University Law Review at SSRN)
[2] Meet Darknet, the hidden, anonymous underbelly of the searchable Web by Brad Chacos (PC World)