The rapid development of artificial intelligence (AI) and systems that act autonomously is constantly leading to new areas of use, business models, and opportunities to add value. The systems in question do not only follow rigid decision trees – they are intelligent, autonomous, proactive, communicative, and mobile. If these autonomous AI systems act independently on the market, there is a risk that agreements entered into by them will be invalid. A solution could be to attribute the declarations made by autonomous systems to their human users.
Deployment of AI in the chemical industry
One of the uses of AI in the chemical industry is to help develop new substances. For example, the algorithms can extrapolate, with high probability, the necessary chemical reactions based on the desired end product. Another group of AI systems can – in a manner comparable to a human – take decisions and interact with others on the market.
For example, they can manage the supply chain by finding the best prices, “negotiating” terms and conditions of purchase, organising delivery, and autonomously managing quality control, including enforcing any defect remediation rights.
Is the legal system prepared for autonomous systems?
AI enables machines acting autonomously to enter into agreements. The question here is: are these agreements legally valid? Are the criteria of corresponding declarations of intent even met by computer programs? When it comes to legal advice, this poses a whole range of new and fascinating questions.
What all types of AI have in common is that they are deployed on the market by their users and are not detectable as machines by those who interact with them. Autonomous systems based on AI have developed to such an extent that they pass the “Turing Test,” which means that humans no longer identify them as machines.
Can autonomous AI validly enter into agreements?
The question of how to deal with agreements entered into by autonomously acting AI systems from a legal perspective has not yet been definitively resolved. Is the AI system a messenger of the user or the latter’s representative? Could an AI system perhaps even be treated as an underage child of the user or even as an electronic legal person in its own right? Strong arguments can be brought to bear against all these views.
Protection of legitimate expectations as a key factor
The deployment of AI systems can probably already be accounted for by existing laws if the perspective of a (human) contractual partner (“objective recipient’s perspective”) is used as a basis.
There are a number of examples of this in legal practice (e.g. blank or computer declaration). As a rule, contractual partners do not know that they are negotiating with machines, but in fact believe that they are interacting with humans. They legitimately expect that these “persons” are entitled to, and have the capacity to, make corresponding declarations. This legitimate expectation is worthy of protection because otherwise legal transactions would become significantly more complicated.
As AI systems can hardly fulfil the agreement and do not possess any liable assets themselves, contractual partners must be able to legitimately expect that users of autonomous AI systems will be bound by the declarations. Businesses that, by using autonomous systems in day-to-day transactions, create the expectation that they are bound by the declarations made by AI must adhere to such declarations.
To that end, if the AI used by a chemical company orders the wrong basic raw materials, an agreement is still entered into between the supplier and the chemical company as the user of the autonomous AI system. The chemical company is bound by this agreement. As artificial as intelligence may be – as a contractual partner, it is very real.
[View source.]
