Two-Year Verdict: ChatGPT’s Impact on the Legal Practice

Banner artwork by Primakov / Shutterstock.com

As ChatGPT celebrates its second birthday this month, it’s remarkable to reflect on how this technology has transformed the legal landscape. When OpenAI released ChatGPT in November 2022, few anticipated its explosive impact — including its own developers.

As Nick Turley, OpenAI’s head of product for ChatGPT, recently revealed, the team had “very tempered expectations” and launched it in a mere 10-day sprint before the holidays. Within days, usage had far exceeded their most optimistic projections.

Where we’ve been

The past two years have seen unprecedented adoption of artificial intelligence (AI) and generative AI (GenAI) tools in legal practice.

ChatGPT became the fastest-growing consumer application in history, reaching 100 million active users within two months. For in-house counsel, some of us began by wading cautiously into experimentation while others dove head first and were sanctioned, causing some lawyers to remain on the sidelines. The past two years have marked a shift from viewing AI as a future possibility to an immediate reality requiring ethical consideration under ABA Model Rule 1.1’s technology competence requirement.

Initial reactions ranged from excitement about increased efficiency to concerns about accuracy and confidentiality. Some legal departments began experimenting with AI for routine tasks like drafting initial communications, summarizing documents, and conducting preliminary research. However, early limitations around accuracy, data privacy, and reliability meant careful human oversight remained essential.

Where we are

Regulatory environment

Recent regulatory developments worldwide are shaping the AI landscape significantly. The European Union continues to lead with its AI Act, which categorizes AI systems by risk levels and imposes strict compliance requirements for high-risk applications. In the United States, the Biden Administration issued an Executive Order on AI in October 2023, focusing on safety, privacy, and equity in AI systems, alongside the NIST AI Risk Management Framework. Globally, countries like China have introduced the Generative AI Measures, emphasizing content regulation and transparency. These frameworks collectively signal a shift toward harmonizing AI governance while addressing ethical and societal concerns.

Adoption and anxieties

ACC’s recently released survey on GenAI in corporate legal work found that GenAI adoption is rapidly accelerating among in-house legal professionals. Nearly a quarter (23 percent) of in-house legal professionals are already leveraging GenAI in their work, and an additional 15 percent are actively exploring its potential.

While concerns about data privacy (49 percent), output quality (45 percent), and company restrictions (29 percent outright prohibit its use for work matters) hinder broader adoption, these figures underscore a swift shift toward AI utilization in legal departments. Notably, only 10 percent of those surveyed say that they are neither using nor planning to use GenAI.

[Explore the data: CLOs Lead the Charge on “Transformative” GenAI Use]

Practical application

Today, in-house legal teams are developing more sophisticated approaches to AI implementation. The technology has evolved significantly, with GPT-4 and subsequent updates bringing improved accuracy, multimodal capabilities (voice, images, and text), and enhanced reasoning abilities.

Lawyers now have a breadth of useful generic AI tools at our fingertips. A search engine like Perplexity can generate precise answers to questions annotated by links to search results (though not without its own controversies). SlidesGPT is just one of a growing number of AI tools that will generate content in PowerPoint format on any topic of your choosing, including legal training. MidJourney, Leonardo.ai, and DALL-E 3 are just a few image generators limited only by your own amazing creativity. Google’s AI-powered research assistant, NotebookLM can:

• Analyze and summarize documents uploaded by users
• Answer questions about uploaded content
• Generate notes and insights from documents
• Maintain citations/sources for its responses
• Help with research by connecting information across multiple documents
• Create study guides and outlines
• Extract key points and themes from text
• Produce an entire podcast of two chatbots discussing the material you uploaded. Yes, believe it!

If you visited the JacksonLewis PC booth at the 2024 ACC Annual Meeting, you could have uploaded a headshot like mine on the left, to generate AI versions of yourself like these.

Legal departments are now using AI tools for myriad applications:

• Document drafting and review
• Contract analysis and management
• Policy and procedure development
• Initial legal research and case law analysis
• Meeting summaries and action item tracking
• Regulatory compliance monitoring

However, successful implementation requires understanding both capabilities and limitations. It continues to be helpful to think of LLMs as a spritely and eager law clerk who has read the entire internet; a highly intellectual know-it-all who is junior in wisdom and experience.

Where we’re going

The next frontier in AI legal tools appears to be moving from simple query-response interactions to more complex task delegation. As Turley explains, we're entering an era where AI can potentially handle longer, more complex tasks independently — from comprehensive contract review to detailed regulatory analysis.

For in-house counsel, several key developments are on the horizon:

• Enhanced integration. AI tools will become more seamlessly integrated with existing legal workflows and enterprise systems. One example is Microsoft Copilot, an AI assistant integrated across Windows, Microsoft 365 apps (like Word, Excel, PowerPoint, Outlook, and Teams), and web browsers. It uses OpenAI’s technology to help with tasks like writing, data analysis, presentations, email management, web searches, and code generation.
• Improved reasoning. New models emphasize thinking before speaking, potentially reducing hallucinations and improving accuracy in legal analysis.
• Specialized legal models. The emergence of legal-specific AI models trained on legal documents and precedents, such as GC.ai offer more precise and reliable assistance specifically for legal tasks. These tools raise critical questions as to how they will shape the legal landscape, and in particular, relationships between in-house and outside counsel.
• Autonomous agents. AI systems may soon handle complex, multi-step legal tasks with minimal human intervention, though always under attorney supervision.

Best practices moving forward

As we enter this next phase, there are steps in-house counsel can consider:

1. Develop clear AI usage policies that address confidentiality and ethical considerations — to the extent you need them. Policies should realistically align with how employees actually use AI tools, as overly restrictive guidelines that don't reflect workplace realities will likely be ignored from the start and could create liability when violations inevitably occur.
2. Implement verification protocols for AI-generated content. Create practical checklists for different types of AI-generated content, with more rigorous verification requirements for high-stakes documents like contracts or regulatory filings, and lighter review for routine communications.
3. Invest in training to maximize the benefits of evolving AI capabilities. Focus on hands-on workshops and real-world use cases rather than theoretical discussions, as practical experience helps team members develop intuition about when and how to effectively leverage AI tools. If you don’t have a budget or bandwidth to spend much time on this, YouTube is awash with prompt engineering courses that offer excellent basics. Classes such as Vanderbilt’s certification course offer systematic but inexpensive education.
4. Stay informed about AI developments and regulatory changes. Assign specific team members to monitor AI developments and regulatory updates in your jurisdiction and industry, ensuring someone owns this increasingly critical knowledge management function.
5. Balance innovation with prudence, particularly regarding sensitive legal matters. Consider creating a tiered system that clearly delineates which matters are appropriate for AI assistance and which require traditional handling, helping team members make quick, confident decisions about tool usage.

The path forward

As we mark ChatGPT’s second birthday, it’s clear that AI has become an integral part of modern legal practice. The challenge for in-house counsel isn't whether to embrace AI, but how to do so responsibly and effectively. By maintaining a balanced approach — leveraging AI’s capabilities while acknowledging its limitations — legal departments can enhance their efficiency and effectiveness while upholding their professional obligations.

The next two years will likely bring even more dramatic changes. The key to success will be developing frameworks that allow legal departments to adapt and thrive in this rapidly evolving landscape while maintaining the highest standards of legal practice.