What is Lexis+ AI?

A generative legal AI solution that adds more value through conversational search, drafting, summarisation and document analysis. Uniquely grounded in Legal Research and Practical Guidance from LexisNexis, results are always linked to verifiable, citable authority or source.
Lexis+ AI is the one legal AI solution to transform your legal workflow.

How is Lexis+ AI different from OpenAI’s ChatGPT?

ChatGPT is not suited for legal research as it does not meet the high standards of accuracy and completeness
required for this work. Lexis+ AI provides authoritative information to users through natural conversations
using large language models. Chat GPT does not have access to this data, and it doesn’t employ techniques like
Retrieval Augmented Generation (RAG) to ground its output in truth.

What is conversational search? How is it different from traditional search and
why is it better?

Our conversational assistant leverages generative AI to engage with users like a trusted colleague to assist in
the completion of a research or drafting task. It uses a neural network architecture to analyse and understand
natural language input and generate appropriate responses based on LexisNexis data. This experience allows for
a dynamic and iterative research process, contrasting with the more linear and potentially frustrating nature of
traditional research.

How does Lexis+ AI guard against hallucinations?

Lexis+ AI is the only legal generative AI solution with citations linked in its responses, providing trusted legal
results backed by verifiable authority. It minimises the risk of invented content, or hallucinations, and checks all
citations to ensure validation. The solution also offers users the ability to input
specific citations to verify accuracy and flag when a citation might be wrong.

How do humans interact with the AI to ensure quality results (fewer
hallucinations, factual errors, etc.)

We employ a “humans-in-the-loop” approach where SMEs, data scientists and engineers are evaluating prompt
responses and making adjustments. This is in addition to the adjustments made based upon direct user feedback,
highlighting a good or bad output within Lexis+ AI. We are using a team of specialty trained legal experts to test
the model’s responses to likely use prompts using a test set of prompts. These experts are trained to evaluate
language model output to ensure consistency and quality.

Book your Lexis+ AI™ demo now

Get ready to revolutionise your legal work with Lexis+ AI^™!
Be among the first to explore this generative AI solution that's set to reshape legal work in the UK.

Will my entries into the tool be used to train Lexis+ AI model?

LexisNexis does not use customer data to tune or train our models. Consistent with the Lexis+ and Lexis service,
we only use customer data that is aggregated and anonymised to improve the customer experience. Users also
have individual control of prompt history and options to delete prompt history from our services. For further
information, you can access our privacy policy.

How have possible responsible AI risks been addressed for Lexis+ AI?

We have a set of Responsible AI Principles that our global teams embed into their working practices.
Responsible AI was a core part of our discovery and evaluation process for Lexis+ AI and we continue to delve
deeper into, and iterate over, responsible AI implications in partnership with our legal, privacy and security
teams.

Will customers with Lexis+ AI be able to search the way it is currently
done on Lexis+?

Yes, the existing capabilities of Lexis+ will be available for use within Lexis+ AI.

Are there recommendations for prompts that will lead to well-developed
answers from Lexis+ AI?

We are exploring methods now for sharing prompt samples, and developing refinement prompts, that will help
train users and lead to greater success when using the system.

How will models update as content is added? Will it be a continuous process?

Lexis+ AI is being developed in a manner to ensure that the output always considers the most recent content
available. The foundational LLM is not being updated on a conti nual basis using our new content, but the latest
content is utilised through our RAG approach to ensure it is considered when developing the output.

Become a Lexis+ AI Insider

Gain exclusive access to insightful webinars, stay at the forefront of legal thought leadership, and be the first to receive crucial updates on upcoming products. Stay ahead of AI developments with timely, relevant, and expertly curated content.

Is the LLM secure without an ability to access open web content ? If it is only
trained on Lexis content, how does it have an understanding when it comes to
building a response (Proper structure, grammar, etc.)?

It is secured as a private model, with all individual user interactions anonymised, encrypted, and not being
utilised as direct feedback to update the foundational LLM. In addition, we provide access to private models
through cloud providers AWS Bedrock and Microsoft Azure, ensuring the privacy and data c onfidentiality our
customers expect. The foundational LLM is not trained only on Lexis content, but also on open web content. Our
system is being fine-tuned utilising LexisNexis content to improve the output, along with prompt engineering to
ensure the desired output structure is followed.

How many conversations can you save and for how long?

On Lexis+ and Lexis+ AI, all user history is available for 90 days and this timeframe will also apply to chat history
from user interaction with the AI model. Customer information is encrypted while in storage and users can delete
chat histories to remove them from their stored user history.

Will there be a way to determine what is copied and pasted from Lexis+ AI
output? How do we avoid plagiarism?

For AI-generated summaries of individual documents, these summaries will be stored and retained with the
document for future viewing. This provides consistency to users, and it also allows you to see exactly what the AI
tool generated, as the content will not change over time. It will be available for any comparisons with the work
product (much like our editorially created summaries for case law). The AI-generated summaries will be marked
as such in the product.

How is the use of Lexis+ AI compared to that of Lexis Answers?

Lexis Answers is a feature of L exis+ that uses Extractive AI to make research more
efficient for our users. Lexis+ AI will in fact utilise our extractive AI capabilities, in combination with new
generative AI capabilities, to accelerate the research process, deliver answers, and research information that will
save users further time compared to today.

How does Lexis+ AI ensure privacy and protect sensitive data inputs?

All Lexis+ AI conversations, including queries input by users, will be encrypted in transit and at rest.
Conversations will be removed from the user’s chat history after 90 days unless the user saves the chat to a
folder where the data could be stored until deleted by the user.

Where does the data go when it's uploaded?

Documents are temporarily processed by our models which are hosted in protected accounts in commercial-grade cloud infrastructure, so we can manage performance and security (see Sub- Processor list - Lexis+ AI™ (file uploads)). Your documents are under your control. Documents can be removed by deleting the session conversation thread or our system will purge them after a 10-minute inactivity. Your documents are retained only for the duration of an active, in-progress session. Your conversations related to the documents uploaded can be deleted by the user or they are purged after 90 days.

Why is data in the US and can it not be in the EU/UK?

Like any large international group, LexisNexis makes technological choices based on various criteria. We select providers which provide optimum customer experience and security. We are listening to our customers and location of model processing will likely change as we evolve our configurations with AWS and Azure.  We will explore alternate processing locations in the future based on input from our clients. 

If data is in the US how are we compliant with the appropriate data transfer rules?

As  per our Privacy Policy, we take steps, including through contracts, intended to ensure that the information continues to be protected wherever it is located in a manner consistent with the standards of protection required under applicable law.

LexisNexis, a division of RELX Inc., has certified to the EU-U.S. Data Privacy Framework (EU-U.S. DPF), the UK Extension to the EU-U.S. DPF, and the Swiss-U.S. Data Privacy Framework as set forth by the U.S. Department of Commerce. Please view its Data Privacy Framework Notice here. To learn more about the Data Privacy Framework program, and to view its certification, please visit https://www.dataprivacyframework.gov.