Contact
Subscribe to the newsletter
Leader in legal information systems
Research
You are here :
  1. Resources
  2. Press Releases
  3. Blog
  4. Towards an Augmented Collaborator: Strategies, Experimentation, and Data Security

Towards an Augmented Collaborator: Strategies, Experimentation, and Data Security

Published on : 27/05/2024 27 May May 05 2024
Blog 

Moise Zapater's intervention, Innovation Director (Septeo), as part of the roundtable "The Impact of AI for Lawyers in 2024: Technology, Skills, Use Cases".

Thanks to AI, data analysis is faster, more relevant, and more efficient, akin to reading a document or observing an image. This is the concept of the "augmented collaborator." This time-saving represents a significant shift as it allows us to reallocate some of our energy to other tasks with higher added value.

The algorithms used in AI have existed for over half a century. However, their exploitation, especially that of generative AIs, only became economically viable on a very large scale since late 2022 with the arrival of the first widely available Large Language Model (LLM) provided for free: OpenAI's ChatGPT.


Revolution of AI: Towards an Augmented Collaborator

According to a study by Amazon Web Services (AWS) and consulting firm consulting firm Strand Partners, the number of companies adopting AI rose by 35% between 2022 and 2023, showing strong adoption in just a few months.

At Septeo, our initial response was to involve the legal department to determine what was feasible and what was not. I believe that lawyers are among the most exposed populations to AI and its use. In fact, they are probably ahead because they are the first ones confronted with it.


Importance of Experimentation in AI Integration


To kick-start the process and persuade others of the added value, experimentation is essential. It's crucial to dive in, to step out of one's comfort zone to truly grasp the stakes. Testing allows for a concrete understanding and measurement of impact. Fortunately, within our programs, we have methodologies to evaluate effectiveness: by equipping one team with the tool while leaving another without, or even by observing productivity improvement within the same team.

The results are tangible: increased productivity, enhanced ability to handle more issues and use cases, as well as an increase in the number of screens managed.

Convinced by these results, the next question is how to protect oneself? Typically, this is the time to consult the legal department.

The first step is to recognize the financial interest for the company, often materialized by productivity gains through these tools. After observing a 20% increase in productivity, the decision to acquire the license becomes evident. However, acquiring the tool raises questions about its integration: how to use it effectively? How to support teams facing what may be perceived as competition?


Overcoming Reservations: The Rise of AI in Enterprises


When we encounter initial resistance from collaborators, especially from the more technical ones, the proposal is to start by testing the tool on non-impactful subjects. Following this, over 80% of the collaborators acknowledge that it changes their lives. The importance of staying attentive to market trends in terms of IT solutions is crucial for us. Initially, we focus on this priority, then proceed with an evaluation.

We quickly obtain tangible evidence of our progress. Six months later, we begin a new budget cycle, where each participant contributes, and we adjust our plans accordingly. This primarily concerns our internal operations. But this process has led us to consider how to leverage this momentum to better serve clients.


Data Security in the AI World: Challenges and Precautions


What actions to take? What challenges to anticipate? The first question is: where do our data go? To the United States? To China?

Today, OpenAI is American, backed by Microsoft. Did you know that you can take two types of contracts, one with Microsoft and one with OpenAI? Both are hosted in the same place. In the first, Microsoft's, it is mentioned "we will not use your data," so you can turn to them if necessary. In the second case, OpenAI's, nothing is written.


Ensuring Relevance and Security of Client Data: A Segmentation and Distinction-Oriented Approach

We conducted a thorough analysis and concluded that it's essential to combine security and relevance. However, a question arises: how to ensure this relevance? It's imperative to precisely leverage the data specific to each client, thereby avoiding any confusion with data from other clients.

This need for distinction ensures the relevance of the provided responses and, as a result, imposes strict data segmentation to address security concerns. This approach isn't a luxury but a functional requirement: The separation of databases becomes unavoidable.
Share on
Publish this article Share on LinkedIn Share on Facebook Send by E-mail

History

<< < 1 2 3 4 5 6 7 ... > >>