Generative AI

A guide for
corporate legal
departments
June 2023
 Introduction
The potential of Generative AI in legal has captured the
imagination in ways unmatched in other professional
services. Given its heavy reliance on precedent and
legislation, and the prevalence of text generation
and review as its core tasks, legal work is particularly
susceptible to AI application. Goldman Sachs estimates
that AI has the potential to automate up to 44% of legal
tasks, a materially higher percentage than any other
profession. GPT-4 has already passed the Bar exam. A
leading provider of Generative AI platforms predicts that
within a year, Generative AI will match the capabilities of
a paralegal, and within five years, it will match those of
an “average lawyer.”

While most industries and corporate functions have

already undergone substantial technology-driven
transformation, legal has remained relatively untouched
due to the unstructured nature of legal data and
the continued reliance on expert professionals. With
the emergence of large language models which
underpin Generative AI, the legal industry has almost
certainly reached its long-awaited tipping point toward
transformative change.

02
Envision a scenario where market. We could see the
Generative AI possesses democratization of legal
the capability to address advice, the transfer of legal
the majority of routine practitioner judgement
legal inquiries, generate and opinion to machines,
highly tailored contracts universal access to justice,
instantly, facilitate market practice replacing
contract negotiations, two party negotiations,
identify contractual risks, AI-based case resolution,
summarize changes in and productivity
legislation and case law, transformation for lawyers.
and even draft legal
arguments. GenAI is Almost every Deloitte
already capable of many Legal conversation with
of these tasks – in a small General Counsel (GCs),
number of years we firmly senior corporate counsel
believe it can be the and Legal COOs over
standard model for doing the past three months
so. has started or ended
with a discussion about
Generative AI will be Generative AI. GCs have
pervasive to the core of an emerging awareness of
what legal services are the remarkable potential
today. So much so that this technology holds.
we also believe it has Now, they are beginning to
the capability not merely ask the obvious question
to automate tasks but - where and how they
to disrupt the entire should start using it?
foundations of the legal

03
What is
Generative AI?
Generative AI is a subdivision of artificial intelligence
that empowers machines to create novel content,
encompassing text, code, voice, images, videos,
processes, and even the 3D structure of proteins.
Though some forms of Generative AI are well
established, it was the large language model (LLM)
underpinning an accessible chat interface that triggered
Generative AI's watershed moment, astonishing even
field specialists. In legal services, text generation
represents the most applicable form of content.

From productivity to legal risk the entire enterprise a real possibility—

It takes time and experimentation to learn provided the right guardrails are in place.
how to use Generative AI as part of legal
workflow to generate meaningful, trusted Applications and outputs
outputs. The temptation is to interrogate When thinking about potential use cases,
Generative AI like a sophisticated search it can be helpful to focus on the different
engine. For models that are in the public ways in which the technology can be
domain, trained on vast reams of historic applied, and the kinds of outputs it can
data from the internet, this often returns generate. Generative AI can be applied
results that are questionable—and in several different ways, all of which are
sometimes irrelevant or inaccurate. More highly relevant to legal services, such as:
specialized Generative AI can however
• Textual content generation: Creating
substantially boost lawyer productivity by
content based on input examples,
automating tasks such as document review
documents, data, or a specific theme.
and comparison, summarization, and
Legal examples include drafting
initial contract or legal memo drafting. By
contracts, memos, policies, advice,
liberating lawyers from these routine tasks,
communications.
AI paves the way for them to focus on
more complex, high-value work, potentially • Classification: Assigning a category or
leading to significant cost savings label to a given input. This could include
and growth opportunities. Of course, the classification of contracts and
Generative AI can do more than just documents for due diligence purposes,
increase lawyer productivity. We anticipate or “tagging” data to create a smart,
that much more sophisticated use cases searchable knowledge management
will be developed, in the area of legal risk database.
management in particular. Generative
• Summarization: Producing a concise
AI’s ability to digest and analyze large
summary of a long text or a collection of
volumes of data from multiple sources
texts. Applications include summarizing
makes continuous audit of legal risk across

04
Generative AI can be due diligence reports, legal research, or
legislative changes.
applied in several different
• Transformation: Converting content
ways, all of which are highly into a new type, format, or style. This
relevant to legal services. could include the simplification of legal
language.

• Q&A: Providing a natural language

answer to a natural language question
based on a text or knowledge base. This
could be employed for legal self-service
requests to answer repetitive questions
that require a level of legal insight.

• Extraction: Deriving specific information

or data points from a given input. A key
use case would be legal due diligence,
or high-volume contract repapering that
requires comparison of content across a
large portfolio of documents.

• Reasoning: Producing logical analysis

given a context or knowledge base. An
example here could be undertaking a
regulatory gap analysis and using the
technology to analyze the impact of the
differences between the two pieces of
legislation.

05
A range of
opportunities
Corporate legal departments should look across the
full range of legal advisory and legal operations work
to identify potential use cases. Deloitte has designed a
Digital Artifact Generation/Validation method to help
innovation leaders determine whether an idea can be
turned into a beneficial use case leveraging Generative
AI. At the core of this method are two of the most
critical elements to consider: the human effort required
to complete a task without Generative AI, and the
necessary effort to validate or fact check the output
from the Generative AI. This leads to a two-dimensional
classification, categorizing use cases based on the
required human effort and the ability of the user to
validate the results. If a task requires effort to execute
but is easy to validate, it might be a good use case.
Examples might include:

• Litigation: Analysis of case data for preliminary antitrust risk assessment

rapid early case assessment, enabling the (reduces the risk of manual assessment
legal department to achieve early case of complex data across multiple
settlement (saving money and time and jurisdictions).
avoiding business disruption associated
• Regulatory compliance: Conducting
with protracted litigation).
regulatory gap analysis (comparing two
• Commercial contracting: Identifying or more areas of legislation/regulation)
the most likely median end position for a and assessing the impact of the
particular clause in a particular contract differences (speeds up a manual and
type to speed up the negotiation process time-consuming process and increases
(helping to deliver business value and accuracy).
increasing speed to market).
• Knowledge management: Generation
• M&A: Using the speed and compute of updates and briefings for the business
power of LLMs to conduct more detailed and the lawyers, generation of training
due diligence across significantly larger materials, interrogation of knowledge
data sets, increasing the accuracy of assets (reduces risk and upskills teams,
target valuations (reduces business risk improves the employee proposition).
and increases business intelligence).
• Legal operations: Automated RFP
• Competition/anti-trust: Gaining a (request for proposal) response scoring
rapid and up-to-date understanding and assessment (eases the burden
of the evolving merger control regimes of a painful process and encourages
across all relevant jurisdictions in the objectivity).
context of a merger and generation of

06
The expanding pool of Options for support: Specialized vs. firm offering a Generative AI bot trained
generic with their expertise is a legal information
Generative AI technology Most lawyers’ experience with Generative provider. From a corporate legal
providers will provide AI to date has been with generic models department perspective, the categorization
freely available online that have not been is less important than the increasing range
a range of service explicitly trained for legal use. While of choices available from an expanding
capabilities for legal these could be used for legal work, the pool of providers that will provide a mix of
real potential lies in creating legal specific technology and legal service capabilities.
departments. capabilities, which could be achieved by
enhancing the data input into generic
models, or by creating dedicated legal
models. We have seen instances in other
industries (e.g., financial services and
healthcare) where industry-specific models
have been developed, and we expect that
this will also happen in legal services.

For corporate legal departments

considering the use of legal-specific
capabilities, several options are available:

• Existing legal tech providers are

incorporating Generative AI into their
platforms; for example, some contract
lifecycle management providers are
incorporating Generative AI into their
platform for contract drafting or
information extraction.

• Some new Generative AI legal tech tools

allow organizations to supplement a
generic (non-Legal) model with their own
proprietary content.

• Legal information providers are

developing Generative AI tools that
provide Generative AI capabilities on top
of their existing knowledge bases.

• Some of these information providers

will allow their clients (law firms and
corporate legal departments) to
supplement the system's knowledge with
their own legal content and expertise.

• Larger law firms and corporate legal

departments may even create their
own capabilities, whether leveraging
legal specific platforms or more generic
Generative AI models.

As the technology becomes more

pervasive, the boundaries between these
categories may blur, raising questions
about whether a legal information provider
offering Generative AI capabilities is
providing legal services, or whether a law

07
Getting
started
General Counsel are increasingly questioning where to
begin. We have identified some key criteria to consider
when evaluating Generative AI use cases, which include:

Demand: • How much human effort is required

to verify the accuracy of the AI result?
• Is there a pressing issue that needs
Generative AI is based on statistics and
addressing? To justify the effort and cost
is not 100% accurate. The outputs from
of implementing Generative AI, corporate
Generative AI therefore require validation
legal departments should focus on a
(see Risks below). If a disproportionate
genuine identified problem that needs
amount of effort is required to validate
solving.
the outputs of the AI, it may make
• Is the volume of work sufficient to justify more sense to continue with the more
a solution? As with any automation, there traditional approach. Good Generative AI
may be incremental gains each time an use cases will be those where the output
activity is performed but real economies can be validated with lower amounts of
of scale are achieved only when the human effort.
activity is performed at a reasonably high
• Who possesses the subject matter
frequency.
expertise to feed into the model? AI
• Does the current work require enough needs to be fed the required subject
human effort to warrant an AI solution? matter expertise to effectively generate
As AI is essentially replacing tasks specialized outputs. When considering
that humans do, a greater return on use cases, consideration needs to be
investment will be achieved if the work given to who has the underlying subject
that is being replaced currently requires a matter expertise to be able to create
reasonable amount of human effort. the best solution. See the section above
regarding Specialized vs. Generic AI.
Feasibility:
• Who is best positioned to provide
• Does the work involve the creation of a support? Should a generic solution be
digital artifact? Generative AI is about used, or a legal-specific one? If so, should
generating digital content. It is therefore an “off the shelf” solution be used, or
key to focus on activities that involve such should something bespoke be created?
content generation.

08
Clear trends are likely To begin answering these questions, the
first task should be to undertake an activity
• Data security: Does the technology/
service provider secure your data to the
to emerge as the use of analysis of the work done in the legal required standards?

Generative AI technology function. This will allow the in-house legal

team to evaluate these activities against
• IP: Is the technology/service provider

evolves over time. the factors listed above to determine which

using your data to train its model for
others’ use?
work should potentially be performed/
augmented by Generative AI. • Obsolescence: Will bespoke solutions
become obsolete when large technology
Once potential use cases have been and service providers release industry
identified, corporate legal departments standard tools?
should consider the best mix of technology
• Expertise: Do you have the appropriate
and services to deliver the use case,
experts to partner with you through this
factoring in the risks set out below. To help
journey?
determine what that mix might be, ask the
following questions:

• Should you subscribe to a standard tool

(whether generic or legal specific)?

• Should you collaborate with some other

corporate legal departments?

• Should you partner with a legal service

provider?

• Should you build something bespoke?

• Should you simply outsource the work?

At this early stage of use of Generative

AI, we believe there may be a mix of
approaches, depending on the specifics of
each organization and use case. Over time,
as the use of the technology evolves, clear
trends are likely to emerge.

Consider the risks

A degree of experimentation is critical to
gain an understanding of the possibilities
of this transformative technology. However,
corporate legal departments need to
understand the risks involved, and work
with the technology safely. When exploring
the use of Generative AI, several risks
require careful consideration:

• Hallucinations: As Generative AI models

operate on the basis of mathematical
probabilities, there is a risk of generating
incorrect outputs. While this risk
diminishes with more domain-specific
models, it is crucial to validate the
model’s output before use.

• Data privacy: Consider what data you

are sharing with the technology/service
provider and where it is hosted.

09
Impact on legal
service providers
Legal service providers (law firms and others) are
exploring the use of Generative AI in a similar
way to corporate legal departments. If the service
enhancements and efficiency gains are as significant
as we expect them to be, the cost of delivering legal
services should significantly decrease. Will law firms
hold on to these margin gains or will they be passed on
to their customers? Several law firms have announced
their use of Generative AI including stating that this will
not result in changes to resourcing models and fees.
We believe that corporate legal departments should be
challenging their service providers on their use of AI and
on the benefits that they will receive.

10
Conclusion
We believe that there has never been a more exciting
time to be working in legal services. We are all on
a learning journey, with the ultimate destination as
yet unclear. As we learn together, we must take this
opportunity to reimagine how legal services can be
delivered in the future. What is clear is that Generative
AI will have a profound effect on multiple facets of
legal services–resourcing models, value chains, costs,
risk management, and service offerings. Given this
transformative impact, General Counsel should be
thinking about this foundational technology in a strategic
manner and positioning their functions to benefit from
its increasing capabilities. They should ask themselves
where it makes sense to apply Generative AI in the legal
function, what it means for resourcing models and skills,
how it may influence sourcing and partnering strategies
and how it could be used to better manage legal risk.
The next few years will bring fundamental change
to the industry, presenting huge opportunity for all
participants.

11
Contacts
Bruce Braude
Deloitte Global Chief Technology Officer - Deloitte Legal
[email protected]
+44 20 7007 2399

Isabel Parker
Partner, Deloitte Legal UK
[email protected]
+44 20 7007 7980

12
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited (“DTTL”),
its global network of member firms, and their related entities (collectively, the
“Deloitte organization”). DTTL (also referred to as “Deloitte Global”) and each
of its member firms and related entities are legally separate and independent
entities, which cannot obligate or bind each other in respect of third parties.
DTTL and each DTTL member firm and related entity is liable only for its
own acts and omissions, and not those of each other. DTTL does not provide
services to clients. Please see www.deloitte.com/about to learn more.

Deloitte Legal means the legal practices of DTTL member firms, their affiliates
or their related entities that provide legal services. The exact nature of these
relationships and provision of legal services differs by jurisdiction, to allow
compliance with local laws and professional regulations. Each Deloitte Legal
practice is legally separate and independent, and cannot obligate any other
Deloitte Legal practice. Each Deloitte Legal practice is liable only for its own
acts and omissions, and not those of other Deloitte Legal practices. For legal,
regulatory and other reasons, not all member firms, their affiliates or their
related entities provide legal services or are associated with Deloitte Legal
practices.

Deloitte provides industry-leading audit and assurance, tax and legal,

consulting, financial advisory, and risk advisory services to nearly 90% of the
Fortune Global 500® and thousands of private companies. Our professionals
deliver measurable and lasting results that help reinforce public trust in capital
markets, enable clients to transform and thrive, and lead the way toward a
stronger economy, a more equitable society, and a sustainable world. Building
on its 175-plus year history, Deloitte spans more than 150 countries and
territories. Learn how Deloitte’s approximately 415,000 people worldwide
make an impact that matters at www.deloitte.com.

This communication contains general information only, and none of Deloitte

Touche Tohmatsu Limited (“DTTL”), its global network of member firms or their
related entities (collectively, the “Deloitte organization”) is, by means of this
communication, rendering professional advice or services. Before making any
decision or taking any action that may affect your finances or your business,
you should consult a qualified professional adviser. No representations,
warranties or undertakings (express or implied) are given as to the accuracy
or completeness of the information in this communication, and none of DTTL,
its member firms, related entities, employees or agents shall be liable or
responsible for any loss or damage whatsoever arising directly or indirectly in
connection with any person relying on this communication. DTTL
and each of its member firms, and their related entities, are legally separate
and independent entities.

© 2023. For information, contact Deloitte Global