The increase in smart systems and fully-automated artificial intelligence in commerce is set to revolutionise society, yet it gives rise to significant questions for legislators as to how to apply the law to systems that are no longer necessarily controlled by people. If machines are capable of decision making by themselves, then how can a wronged party be compensated if there is no wrongdoer, and can a smart system without a legal mind be held accountable for any such wrong caused?
We have previously looked at how blockchain technology can automate commerce on a mass scale, particularly within supply side solutions. The potential for purely automated ordering and shipping solutions is now a reality: smart factories that can read customer data to know what stock to order and when.
Henry Ford’s Model T car factory back in 1904 was a smart system, albeit relying on humans to operate the conveyer belts. Car factories have since replaced many of those humans with robotic arms and fully automated manufacturing processes. These are smart systems in themselves – but now smart factories can go much further than this. With the internet, factories can expand their networks outside their own 4 walls and link up with other factories and businesses globally to talk to one another and share data. Not only can warehouse operations become fully automated and self-sufficient, but also things outside of the factory, such as inventory tracking, supply and demand levels, logistics and consumer wants and complaints. These are all things that can be logged and the data used to improve the supply chain and end user experience. A decade ago people where weighing up the benefits between having to visit a physical shop to do the shopping compared to ordering online through the shop’s website; soon the question will be whether we should even bother thinking about shopping at all, or just have the shop decide for you.
In 2016 India’s Prime Minister Modi launched its new smart port initiative for its 1600km coastline, and other countries are also incorporating them. Smart Ports use technology to load and move goods, fulfil orders and place containers over the rail all through automation. Self-driving vehicles are being put to use in other areas too: all of this is converging toward a purely automated process for B2B commerce in general.
At B2C level, Samsung has spoken of smart fridges for a number of years. They are now available to buy. At first, a fridge connected to the internet seems excessive. But then, if manufacturers and suppliers are all connected to the same network as your fridge, you can see the benefits. Orders for food can be placed when the fridge is near empty; faults and maintenance can be identified and logged to prevent faulty products; even dieting can be programmed.
How is such automation possible?
It is all possible through connectivity and the transfer of data. Data has become a commodity beyond something that can just be sold to advertisers to target consumers. The mass amount of data that is now readily available and transferred could easily allow a fully automated bot to make smart business decisions based upon that data.
“Data has become a commodity.”
With this brings massive implications at law and raises questions as to how to regulate and control this new era of commerce. There is a need for modern interpretations and new law to deal with this new way of life.
The systems automation is smart to pre-empt the next step in a supply chain, but what they cannot do (yet) is use discretion like a human can, or deal with situations which are out of the ordinary. This could result in some unusual or even unjust results. For example, a seemingly smart decision for a business not to fulfill a certain order could lead to delays in a logistic’s supply chain, which leaves a retailer at a loss.
The impact of Artificial Intelligence on legal remedies and how to deal with machines that make decisions without input from a legal person.
There is a thought provoking consultation on the UK IPO government website considering the implications of artificial intelligence on intellectual property rights.
The Adapting Law
The law is constantly able to adapt. However, this time it is slightly different …
As an example, the uk.gov article looks at the impact Ai could have on trademark infringement.Section 10 of the Trade Marks Act (TMA) 1994 defines trade mark infringement as:
10 Infringement of registered trade mark.
(1)A person infringes a registered trade mark if he uses in the course of trade a sign which is identical with the trade mark in relation to goods or services which are identical with those for which it is registered.
The difficulty the law will face in adapting to artificial intelligence, as discussed in the consultation through the area of trademark law, is that Ai is not a person. Laws have always been enforced against people or groups of people (countries) or organisations controlled by people (companies). Never before has the law had to deal with the issue of enforcement against machines. Up until now, machines have always been controlled by people, but Ai strives for machines to think for themselves, so it brings a completely new challenge for the law and the way in which our society thinks about governance.
“There is a need for modern interpretations and new law to deal with this new way of life.”
Pursuant to TMA section 10, it is a person who infringes a registered trademark by its actions. By that definition a machine cannot infringe. Currently machines can implement decision making by people, but if the supply side becomes fully automated and Ai begins to make smart decisions on behalf of people, it is no longer a person making those decisions or inputting its instructions to the computer to be carried out. How can infringements, such as supplies to a wrong territory, be policed or how can infringements be remedied if they are not caused by the fault of another. With fully automated smart solutions, there could be a wronged party, without a wrongdoer.
To put it another way, in trademark terms, if goods were being imported for sale into a certain market and a party held exclusive rights to the IP over that product in a particular market, what would happen if an Ai process made a decision to import the wrong product into the market thereby infringing the rights holder’s IP. Would it be possible to enforce a finding of infringement against a person who is not culpable or involved in the decision that led to such infringement?
Gradual Shifts Rather than Overnight Change
It is likely a far way away from the point where Ai machines are intelligent enough to make independent decisions without the input of humans, but one day it will happen.
Perhaps the gradual shift in our advancing civilisation is the reason that the law is able to continuously adapt and remain relevant. After all the overriding objective of the court is to deal with cases justly, and rules on the interpretation of statute allows Judges to be pragmatic when seeking to interpret statute in a way that gives effect to the original intention of the legislator at the time of drafting, and to protect against the mischief for which the statute was created.
The advancement in artificial intelligence will likely see new insurance policies created to deal with the new risks from rogue automation systems that contravene the law and result in liability for the ultimate owners of the products or the machines themselves. The technology may add to our increasingly litigious society. There are already data protection policies to protect against hacks and viruses which did not exist 20 years ago, so arguably the risk of liability caused by artificial intelligence is nothing new and it will simply become another area of law to protect against as society continues to advance.
Arguably the issue of how to identify a perpetrator when decision making involves coded machines, is no different to the issues faced by the advent of the company. Salomon v Salomon (1899) is the authority case finding that a company has its own legal fiction and is a “person” in itself. For a person at law does not need to be a human, it just needs to be anything that has a legal fiction. A company is a “person” for the purpose of law, and we know companies have controlling minds behind it being the board of directors. If a company commits wrong, the law can be enforced against the organisation itself to make good any damage caused in the same way as it is enforced against an individual.
Machines are coded to operate in a certain way. They have rules that they must abide by and those rules are created by humans. Intelligent learning by machines in everyday life is not yet a reality but it still creates thought-provoking questions on how the law will be able to control these intelligent machines which must one day be answered, but they are not questions that necessarily need to be answered right now.