Can AI replace lawyers in International Arbitration? Or be used in International Arbitration - A [guided] interview of a generative AI

This article was first published on LinkedIn on 30 June 2023 (see here)

I have finally decided to use a generative AI (thanks to Bing GPT) to generate a memo.

Spoiler: My conclusion is that It was like conducting a cross-examination of a recalcitrant witness; you get more or less what you want, but in pain. It would be desirable for the AI to be programmed to respond as in a direct examination.

I chose three of my favourite AI / IA related topics (already on my website that the AI decided to ignore!):

• Can AI cannot replace human beings as lawyers and more specifically as counsel for the parties and as arbitrators in international arbitration cases?
• Flaws that affect AI?
• How can AI be used to enhance efficiency and reduce costs in international arbitration cases?

To obtain enough substance presented in an orderly manner, distinct requests were made to cover main topics separately; another request was made for a merger of the two memos; plus, another request for the addition of the last topic. Specific instructions were given for the insertion of titles and sources.

Here is the AI generated memo (final version after merger of prior memos and requests for additions) that is the result of the "AI interview" I did:

================

AI Limitations in Law

Artificial Intelligence (AI) technologies, especially machine learning and natural-language processing, are already impacting and gaining traction within the legal sector. With the development of technology in recent years, AI is changing the way lawyers think, the way they do business, and the way they interact with clients 1. Experts believe the time is ripe to also seriously explore the technology’s use throughout the entire arbitration process in order to, among other things, enhance efficiency and reduce costs 1.

However, AI still has limitations when it comes to replacing human beings as lawyers, counsel for the parties, and arbitrators in international arbitration cases.

AI is not yet able to replicate the services of a lawyer or judge

AI is not yet able to compose legal documents, advise clients, or replicate the services of a lawyer inside of a courtroom. Since AI is dependent on human directives, inputted data, and incapable of certain types of critical thinking, AI is limited by the way in which it is built 2. The use of AI in judicial systems is being explored by judiciaries, prosecution services, and other domain-specific judicial bodies around the world. For example, in the criminal justice field, the use of AI systems for providing investigative assistance and automating decision-making processes is already in place in many judicial systems across the world. However, AI is not yet able to replicate the services of a judge or arbitral tribunal.

Flaws that affect AI

There are several flaws that affect AI:

• Hallucinations: In artificial intelligence (AI), a hallucination or artificial hallucination (also called confabulation or delusion) is a confident response by an AI that does not seem to be justified by its training data. Hallucination can occur when the AI model generates output that is not supported by any known facts. This can happen due to errors or inadequacies in the training data or biases in the model itself.

• Data drift: Data drift occurs when there are changes in the distribution of input data over time. This can cause a decline in a model’s ability to make accurate predictions due to changes in the environment in which it is being used 3.

• Biases: Biases can occur when there are errors or inadequacies in the training data or biases in the model itself. For example, if a machine learning model was trained on biased data, it may produce biased output.

• Inability to think outside the box: A big disadvantage of AI is that it cannot learn to think outside the box. AI is capable of learning over time with pre-fed data and past experiences but cannot be creative in its approach 4.

Reasons why AI cannot render justice in lieu of human beings

There are several reasons why AI cannot fully replace human beings to render justice:

1. Lack of critical thinking: AI is incapable of certain types of critical thinking that are necessary for legal work. This means that it cannot fully replicate the services of a lawyer or judge.

2. Dependence on human directives: AI is dependent on human directives and inputted data. This means that it can only operate within the parameters set by humans.

3. Inability to compose legal documents: AI is not yet able to compose legal documents or advise clients. This means that it cannot fully replace the services of a lawyer.

4. Inability to replicate courtroom services: AI is not yet able to replicate the services of a lawyer inside of a courtroom. This means that it cannot fully replace the services of a lawyer or judge.

5. Flaws such as hallucinations and data drift: AI is affected by flaws such as hallucinations and data drift. These flaws can cause errors in its output and limit its effectiveness.

6. AI-based machines don’t have emotional intellect: Emotional Intellect is what we can be proud of. We can express emotions and understand them in others. This ability helps us to build relationships, communicate effectively, and empathize with others 1.

7. AI lacks human-like consciousness: AI systems are not capable of experiencing consciousness in the same way that humans do. They do not have self-awareness or the ability to experience emotions 1.

8. AI has limitations in adaptability: While AI systems are capable of learning and adapting to new situations, they are not as flexible as humans when it comes to dealing with unexpected or complex situations 1.

9. AI requires human oversight for ethical reasons: There are concerns about the ethical implications of using AI systems to make decisions that affect people’s lives. Human oversight is necessary to ensure that these systems are used in an ethical and responsible manner 1.

10. AI may not be able to fully understand and interpret the nuances of human behavior and emotions: These nuances are often crucial in legal decision-making.

How can AI be used to enhance efficiency and reduce costs in international arbitration cases?

Despite its limitations, AI has great potential to enhance efficiency and reduce costs in international arbitration cases 1. For example:

• Data-driven elements: AI can be used for data-driven elements of international arbitration such as e-discovery, proofreading, translation and legal research 4.

• Selecting suitable experts, counsels, and arbitrators: There are already existing tools such as Arbitrator Intelligence which use AI for selecting suitable experts, counsels, and arbitrators 1.

• Facilitating procedural automation: AI can be used for facilitating procedural automation by translating, transcribing, summarizing evidence, and even drafting compilatory parts of legal documents and arbitral awards 1.

In conclusion, while AI has made significant advancements in recent years and has great potential to enhance efficiency and reduce costs in international arbitration cases 1, it still has limitations when it comes to replacing human beings as lawyers, counsel for the parties, and arbitrators in international arbitration cases.

Sources: 2: National Law Review 1: Example 2 [i.e., one of the memo generated] 3: DataCamp 4: Kluwer Arbitration Blog

En savoir plus [suggestions made by Bing GPT]: 1. datascientest.com2. datacamp.com3. builtin.com4. simplilearn.com5. forbes.com6. technologyreview.com 

================

Conclusions:

• Several additional requests and some input and suggestions were needed to obtain more substance on the text dealing with AI flaws, and in particular the notion of data drift, the lack of critical thinking that I had to suggest (as a question, in an attempt not to force the AI to give a positive answer).

What does that mean? Not enough material in the database, or the database is dated, or the AI is flawed or biased not to address certain issues? Does it add some specifically requested items to please the user, or is this the consequence of an hallucination? None of the above?

• At the end, the AI generates a reasonably decent memo but the style is not attractive.
• The first text generated is poor and lacks substance (may be due to a lack of instructions).
• The term used is "AI", with no definition given of what this is. No mention is made of Generative AI, or of General AI Also (it is true that I did not make a specific request) .
• The generation of content requires that one has enough knowledge of the subject matter.

In my opinion, the generation of text by a generative AI is not a complete waste of time but I do not think that there was any gain of time either. Not to mention the need to revise the text in its entirety (which I did not do).

My conclusion is that It was like conducting a cross-examination of a recalcitrant witness; you get more or less what you want, but in pain. It would be desirable for the AI to be programmed to respond as in a direct examination.

Maybe the correct conclusion is that I do not know how to use generative AI and that I need to become an "augmented lawyer" with my own General AI assistant to instruct a Generative AI in my place.