6 Best Practices for Integrating AI into Legal Workflows
Consider two lawyers preparing for a complex commercial matter. Both have access to ChatGPT or Claude. Michael prompts a generative artificial intelligence (AI) platform for legal summaries, but receives generic responses filled with American concepts and propositions. He eventually abandons the tool, frustrated. Meanwhile, Gerry approaches the same task differently. She first considers which aspects of her preparation are suitable for AI assistance, prompts a tool with a clear Australian legal context, critically evaluates each output, and maintains detailed records of AI contributions. Her preparation is thorough, efficient, and ethically sound.
The difference? AI fluency.
While the previous chapter explained what AI is, this chapter addresses a more pressing question: how do we work effectively, efficiently, ethically, and safely with AI in legal practice? Technical knowledge alone is not enough. Just as knowing how a legal database works does not make you a skilled legal researcher. Therefore, understanding AI technology does not automatically translate to productive AI collaboration. This chapter introduces the AI Fluency Framework for Australian legal practice. It is a structured approach to developing meaningful collaboration with AI systems, with a focus on generative AI. By developing key AI competencies, you’ll transition from basic AI use to strategic and responsible integration that enhances your practice while upholding professional standards.
What is AI Fluency?
AI fluency represents a fundamental shift from merely knowing about AI to actively collaborating with it. It is the “ability to work with AI systems in ways that are effective, efficient, ethical, and safe.”[1] More specifically:
- Effective: Achieving your intended outcomes
- Efficient: Optimising time and resources appropriately
- Ethical: Maintaining professional obligations and values
- Safe: Protecting client interests and confidential information
Therefore, AI fluency in the context of Australian legal practice is the confidence and competence to understand, evaluate, apply, and continually govern AI tools so that the resulting work product remains accurate, efficient and ethically compliant (for example, with the Legal Profession Uniform Law, the Australian Solicitors’ Conduct Rules, court directives and client expectations).
Three Modes of AI Collaboration in Law
Integrating generative AI into legal workflows requires careful consideration of when and how to leverage AI assistance effectively. Not every task or project may be suitable for AI involvement, and legal professionals must exercise judgment in determining the appropriate level of AI integration.
To help navigate interactions with AI assistance, Darkan and Feller identify three modalities of interaction with current generative AI.[2]
Modality 1: Automation (AI Performs Human-Defined Task)
- AI performs tasks independently, but based on direct human instructions (e.g. in response to a prompt).
- This modality is particularly useful for improving the efficiency of repetitive, time-consuming, or data-intensive tasks.
- Requires clear task definition and quality control measures.
- Examples: Emails, summaries, social media posts, basic coding.
Modality 2: Augmentation (AI and Human Perform Task Collaboratively)
- AI and human co-define and co-execute tasks in an iterative way, collaborating towards an end goal
- This modality focuses on enhancing human creativity rather than replacing it through the addition of an AI thinking partner.
- Involves a dynamic interplay between human and AI contribution.
- Examples: Writing stories, essays, research papers, complex coding tasks.
Modality 3: Agency (Human Configures AI to Perform Tasks Independently)
- Human configures AI to independently perform future tasks (including for others) on behalf of the user.
- This modality defines the characteristics and future behavior of an AI, rather than a specific task.
- Requires sophisticated understanding of AI capabilities and limitations.
- Examples: Interactive game characters, tutors, chatbots.
Applying Dakan & Feller’s modalities, the ways a lawyer may interact with AI include:
1. AI as an Assistant, where AI performs defined legal tasks. These tasks may include:
- Document review
- Contract analysis and extracting key dates or terms from clauses
- Legal research and case law summaries
- Formatting citations to AGLC4 standards
- First-draft preparation
2. AI as a Collaborator, where a lawyer and AI work together on tasks. These tasks may include:
- Brainstorming negotiation strategies
- Developing legal arguments through iterative refinement
- Multi-jurisdictional comparative analysis
- Analysing complex fact patterns
3. AI as a Representative, where AI acts on behalf of a lawyer. Such situations could include:
- Client intake chatbots with escalation protocols
- Automated compliance monitoring and conflict checking systems
- Document assembly tools for standard agreements
- Legal information services for self-represented litigants
Although this provides several examples of how a practitioner may interact with and use AI. Interactions with AI can span multiple modalities.[3] Practitioners may frequently cross between modalities when the context requires it, even within a single project or workflow.
