Why is legal content ideal for AI?

There are many types of legal documents (laws and provisions, judicial decisions, articles – to name just a few) – they are both versatile and diverse, they deal with many different issues, are used for many different purposes, etc. Yet all of them have something very important in common. One obvious thing would be the subject of law, but this isn't the most relevant familiarity they share in this particular case.

Example 1:
A legal provision
Example 2:
A transcript of a witness's hearing

On a highly abstract level they share several structural similarities – in informational terms: a shared pattern can be observed, or several patterns for that matter.

The origin of this pattern is the legal norm itself, so let’s call it a norm pattern. We know that the legal norm consists of two basic elements, although the second one is optional:

                                1.   The disposition – abstractly defines one or more legally relevant facts that can occur in a society and gives them legally relevant value. The disposition can be a pure definition of a legal term (e.g. a general principle) intended for general (or specific) use when interpreting social relevancies. But the disposition can also be a definition of a certain legally relevant value that needs to be enforced by a sanction in order to guard an important social value by preserving the basic principles of law and legality in a society;

                                2.   The sanction – defines the consequences that must follow whenever undisputable facts exist that correspond with the disposition that needs to be enforced by sanctions.

Example 3:
The disposition and the sanction.
Example 4:
The recognition of relevant facts.

We could say that the disposition consists of – or better: is based upon – certain existing or anticipated facts, yet its definition is of such abstract quality that these facts can no longer be deduced logically once the disposition has been set, but when encountered in an actual situation a clear and direct association between the abstract norm and the concrete facts can be established, thus uncovering the third element of the norm pattern:

                                3.     The association between the norm and the facts – the logical link between the relevant facts recognized in evidence on one hand and the disposition of the norm on the other.

Example 5: The association between the facts and the legal provision.

This third element - although central in its relevance to the case as a whole - is only ever relevant to the norm when the norm itself is actually used: first to identify relevant circumstances, then to execute the sanction.

These are three basic elements of the pattern found in any legal content: the disposition, the sanction, and the association with the facts (or at least the possibility of one). While in the text of laws and statutes only the first and the second element are plainly obvious, in judicial decisions and articles the third element is quite frequently the most obvious one.

A judicial decision is (in informational terms) an application of the norm on the facts emanating from one or more concrete circumstances:

                                  I.   The disposition is associated with the facts – thus identifying the relevant norm and assigning the facts appropriate legal value;

                               II.    The sanction (if applicable) is executed – identifying and applying relevant legal consequences.

When the sanction is executed another element of the pattern becomes obvious:

                                4.   The effect of the consequences – the change in facts and circumstances alike that originates from the application of the consequences when the sanction is executed.

All four elements can be found both in judicial decisions and in articles written on specific legal issues. We could even say, that some articles represent decisions made in hypothetical circumstances compared to a judicial decision which is made in actual circumstances.

Once we’ve identified all four elements of the pattern the informational essence of legal content becomes quite obvious, and leads quite naturally to many interesting possibilities:

                                a)    Improved ability to acquire and recognise legal content;

                                b)   Improved ability to associate individual units of legal content – to group legal documents in accordance with relevant criteria;

                                c)   Improved ability to derive new information from existing content and to create new units of content which in turn can even lead to new derivations;

                               d)   Improved ability to predict future development of legal information, and even to guide it.

The possibilities mentioned above are clear improvements of legal information management, but there are also possibilities that can elevate content management to a new (and currently insufficiently explored, even unexplored) level: the ability to implement methods of artificial intelligence into individual processes of legal content management.

The very elementary implementation of AI in legal content would be:

                                    -     To identify the elements of the norm pattern in order to aid in the presentation of the content – the results of improved content recognition could be made more obvious to the reader, helping him better understand the essence of the matter by obfuscating irrelevant data and pointing out the truly relevant facts on one hand, and help the reader identify corresponding laws and statutes on the other. These constitute the basis of decisions, while the former provide an illustrative context which also has its own value in decision making.

Example 6:
The improved recognition of relevance.
Example 7:
Legally relevant facts within the relevant context.
Example 8: The improved association between the norm and the facts.

This elementary implementation of AI in legal content management improves the usability of content, but at the same time it sets firm foundations for possibile extensions of a more complex, more powerful, and possibly even more useful nature:

                                    -     Machine assisted decisions – a natural upgrade of the improved presentation of content is the automatic creation of suggestions – guides to the final decision which is still made by the judge autonomously. Basically, this means the possibility to automatically create new content – documents that have partially prepared content based on the facts and norms used in a particular case, prepared documents that the judge can use when reaching a decision. In no way shall the judge be bound in his tasks to use any automatically generated content at all – the purpose of the content is merely to assist the judge in recognising relevant content;

                                    -     Automatic decisions – a very bold upgrade of machine-assisted decisions would be to reach the final decision fully automatically. This could be observed as a natural upgrade to the previous level of machine autonomy: based on one or more automatically created possibilities the most appropriate is either selected from the existing ones or a new one is created based on the existing suggestions. Possible implications, however, put a big question mark on such an upgrade. At the moment automatic decisions could only be possible in extremely straight forward cases, but never without proper supervision. At present human intervention in judicial decisions remains essential and completely unsurpassable.

The possibilities of implementing artificial intelligence in legal content management are clearly identifiable and the development of usable solutions highly plausible. Theese possibilities will provide a starting point for many improvements both in legal content management and in the general area of content management, and they also open doors to even more revolutionary possibilities that could find use wherever AI is used to either support human decisions or replace the human in decision making altogether.

© 2005–2006, Mi Lambda, Matija Lah, s.p. All rights in this document reserved.