BC Law AI News & Insights: May 2025 Edition
Welcome to the Administrative and Technology Resources Department’s inaugural AI newsletter. Our goal is to keep the BC Law community apprised of developments at BC and elsewhere in this rapidly changing field and to highlight resources that support the informed, responsible use of these tools.
Before we dive in, taking place next week on May 14th is BC’s first artificial intelligence conference called “BC Talks AI”. If you missed your chance to register, don’t worry! It’s not too late to attend. You can find out more about the event at the conference program website. The conference organizers have asked that you fill out this form to indicate your interest in specific sessions to help with planning purposes
🎥 AI Training: Effective Communication and Collaboration with AI Tools
On April 8, 2025, Kyle Fidalgo led a workshop on effective techniques for prompting and collaborating with AI. You can view the recording here: Effective techniques for prompting and partnering with AI.
Topics include:
- Structuring your requests (Goal, Expectations, Context, Resources)
- Providing high-quality context and source material
- Showing vs. Telling: Using examples (few-shot prompting) for better outputs
- Meta-prompting: Asking AI to help you craft better prompts
- Adopting a collaborative mindset for working with AI assistants
Whether you’re drafting policies, writing emails, or brainstorming ideas, this guide will help you get better results, faster.
🚀 Tool Spotlight: Google Gemini for BC Users
Access Google’s powerful generative AI chatbot, Gemini, free through your BC Google account. Your privacy is protected: inputs via BC’s access are not used for model training.
Why use Gemini?
- Access to the latest Gemini models: Choose between 2.0 Flash for fast, everyday help, 2.5 Pro (experimental) for complex tasks, and Deep Research mode for long-form, in-depth exploration. These models have achieved high scores on recent AI benchmarks and are highly capable of assisting with a range of tasks.
- Data privacy and protection: BC’s Gemini access includes data protection. That is, when accessed through your BC Google account, Gemini will not use your inputs to train its models. Nevertheless, you should not enter ‘confidential’ or ‘strictly confidential’ data, as defined by the BC Data Security Policy, in this or any other online AI tool.
- Create custom AI assistants (Gems): Design your own helpers with specific instructions and workflows. Think of them as your personalized AI co-pilots. Check out the Gems feature.
- Large Context Window: A context window of 1 million tokens allows for sustained, detailed conversations and the ability to reference long documents.
- Image generation: Generate images directly in chats using Imagen 3 (in Flash mode).
Learn how to access Gemini with your BC account.
🧠 New AI Tools Enhance Legal Work Efficiency and Quality, Study Finds
Does the latest AI help lawyers? A recent randomized controlled trial, the first of its kind, provides early evidence that AI reasoning models, which structure complex reasoning before generating output, and Retrieval Augmented Generation (RAG), which grounds AI-powered analysis in legal sources, can, in certain cases, enhance the efficiency and quality of legal work.
Key Findings:
- Efficiency: Participants using both AI tools completed tasks considerably faster, demonstrating productivity increases ranging from 34% to 140%, depending on the specific tool and task.
- Quality of Work Product: Work product quality improved across several assignments compared to the no-AI control group. Both tools led to gains in professionalism, clarity, and organization. The reasoning model (o1-preview) also produced statistically significant improvements in the depth of legal analysis for three of the six tasks tested.
- Accuracy: Overall accuracy results were mixed, with the study finding no consistent improvement from either AI tool across tasks. The RAG tool (Vincent AI) produced a similar rate of factual errors or “hallucinations” as the group working without AI, while the reasoning model (o1-preview) showed improved accuracy in a specific single-document analysis task.
Other News & Model Releases of Interest
AI Lab News & Model Releases
- OpenAI announcements
- Google Announcements
- Anthropic Announcements
Interesting Articles
- NYTimes: Teachers Worry About Students Using A.I. But They Love It for Themselves
- The Cybernetic Teammate: Having AI on your team can increase performance, provide expertise, and improve your experience
- Every: Your CEO Just Said ‘use AI or else.’ Here’s What to Do Next. Learn AI Like It’s Your Job
- A Judge’s Insights: A call for education over regulation
✈️ Field Notes: ABA Tech Show Takeaways
Attending the ABA Tech Show reinforced a clear message: now is the time to gain AI competency.
Here are a few highlights:
- AI and Professional Responsibility: Ethical and professional frameworks are already in place. ABA Formal Opinion 512 and the Model Rules of Professional Conduct (e.g., 1.1, 1.6, and 2.1) provide the structure; now it’s up to legal professionals to skillfully apply them.
- Evolving Client Expectations: Discussions noted increasing client awareness of AI tools and potential expectations for their responsible and strategic deployment in legal services.
- Acknowledging AI Limitations: Concerns were raised regarding AI “hallucinations” and other potential inaccuracies, highlighting the need for human oversight, training, and validation processes.
- Measured Adoption: Recommendations favored careful, phased implementation through focused pilot programs. In short, the advice is “Don’t roll out AI tools with a megaphone!” Instead, start with focused pilots, create small wins, and build champions who can carry the message forward.
- Effective Interaction: Using AI effectively was described as requiring thoughtful engagement and iterative skill development rather than relying on simplistic prompting.
A longer blog post with my full notes and reflections will be posted soon on the Ed Tech Blog. Stay tuned!
Have questions or ideas? Want help creating your own AI workflows? Reach out to Kyle Fidalgo @ atrinbox@bc.edu.
Ready to build your AI competency? Discover AI literacy resources at AI Foundations.