Generative AI for chief legal officers: How to get started, increase ROI and manage risks

Five GenAI guidelines for chief legal officers

1. Understand what’s different

With conventional AI, each new task usually requires you to build a new model from scratch — or “retrain” an old model, which is often costly and time consuming. But GenAI models come “pre-trained” on billions of data points. That helps make these models faster to stand up. It also helps make them more easily adaptable to new tasks, since they draw on such a vast knowledge base.

2. Build trust and address risks

As a legal professional, you may feel that being risk averse is part of your job description — and letting AI read or even write your documents may sound risky. At PwC, we understand. Our brand is built on trust: our clients entrust us with their data and proprietary processes. We’re also highly regulated. Our tolerance for legal violations, compliance issues and data breaches is zero. We’re only deploying AI at scale, with the right governance, because it can also build trust.

Trustworthy GenAI begins with licensing a private version of a GenAI model, then deploying it within a safeguarded IT environment. For us, safeguarding means that the GenAI model is walled off from the outside world. With rigorous information governance, you can then feel safe augmenting the private GenAI model by incorporating or “embedding” your proprietary data — such as in-house examples of briefs. Your legal professionals can review the model’s outputs, treating them like first drafts. This skilled human oversight can put guardrails around potential faults and risks. It can also further refine the GenAI model to better meet your needs.

To help guide your people and provide governance, you’ll likely need responsible AI practices. Some responsible AI practices are for your legal professionals (and leadership) to apply. Other practices are for specialized AI governance roles that complement broader information governance and security teams. The need for trust in GenAI may also provide an opportunity for the legal function to play a bigger role throughout the organization: You can help manage legal and compliance risks related to the use of GenAI in other functions too.

3. Prepare your workforce

GenAI is not here to “take away people’s jobs.” But it can change those jobs and allow redeployment — for the better — if you prepare correctly. By gathering, analyzing, and providing easy access to data, GenAI can vastly speed up legal research. When GenAI writes first drafts of common documents, such as subpoena responses or contracts, your legal professionals can focus on reviewing and improving them. That can spare them tedious work, allowing them to focus on strategy and complex legal issues. It can also grow capacity.

But your team should know when to use (and not use) GenAI; how to craft thoughtful “prompts” that can result in high quality outputs; how to review those outputs and refine the GenAI model by incorporating relevant data to achieve better results; and how to apply responsible AI practices for trusted, compliant and ethical use. Lawyers can often be highly skilled at writing prompts. It’s a task that aligns with their training to be intentional and precise in words and questions.

The goal should be to build a function that is tech-powered, but human-led. People will still make the big decisions and perform the strategic, complex work, but technology can help along the way.

4. Think big

GenAI will likely reinvent the legal profession. But it won’t do it tomorrow. Among other limitations, there isn’t enough computing power right now to deploy GenAI everywhere. But it’s still important to prepare a strategy that doesn’t focus on individual use cases. That would be missing the forest for the trees. Instead, your strategy should consider GenAI’s incredible scalability: how a single model can, with a little refinement, work across your function. 

Given GenAI’s wide applicability, your GenAI strategy should also be part of a larger legal organizational rethink — such as many chief legal officers have already begun doing. Some big picture questions to consider include:

• What new value can your paralegals and junior lawyers deliver, when GenAI accelerates both legal research and first drafts of legal documents?
• What work could GenAI help you do in house, reducing the need for outside counsel? 
• What innovation can legal technology teams deliver, as GenAI speeds up software development? 
• What new value can you provide the business when GenAI is an automated assistant (providing the right insights in the right format at the right time) for your senior people? 

5. Integrate into your tech ecosystem

If your legal function is like many, you may already have a lot of tech — both home grown and from third parties. These tools typically operate in silos: they perform different tasks, for different people, on different systems, using different documents and data sets. Document management, patent management, matter management, contract management, legal holds, e-Discovery, entity management, deal rooms and more are typically systems that don’t talk to each other. 

These silos can stop GenAI from accessing the data it needs to scale and perform as it should. It could be hard, for example, to train GenAI on examples from your more experienced lawyers’ work, if this work is on different, isolated systems. That’s why, as you prepare to integrate GenAI into your legal technology ecosystem — where it can help almost every tool work better — you should start or accelerate work on connecting these tools and organizing your data. 

GenAI can also help accelerate legal tech integration. Similar to how reporting can work, GenAI can sit across and tap into your systems. Its user-friendly interface and its ability to make sense of unstructured data can, over time, help your other technology tools access more thorough and reliable data. That can also reduce the need for “rip and replace” tech efforts, enable more rationalization, and increase your ability to turn data into insights and value to the business.

What to do now

With GenAI, the place to start is trust, which requires security, governance and training. Start by making sure that you have a safeguarded IT environment. License a private version of your chosen GenAI model. Negotiate terms with the vendor and set up data governance to help safeguard your data to keep it private. Set policies for how your people can use GenAI: typically only within your safeguarded environment, and only for tasks for which they’ve been trained to use GenAI. Establish clear governance guidelines and teach your team how to use GenAI responsibly.

With a trusted foundation established, it’s usually leading practice — in close collaboration with both legal tech professionals and data scientists — to take advantage of how GenAI’s remarkable scalability can lead to significant value quickly. If you accelerate efforts to clean up and connect your data and technology ecosystems, GenAI can likely deliver even more value quickly. Also important is a plan to manage the risks that GenAI can bring.

Once you get started with GenAI, capacity growth and lower costs can come quickly. GenAI is especially strong at supporting knowledge work based on well-defined rules, given in precise language, which lawyers are experienced at crafting. The legal function may, in other words, be especially well positioned for GenAI. With a strategic approach, the legal function could find itself not only delivering more value more quickly to the business, but also showing leadership in the organization’s transformation.