How are people using AI to address their legal problems, and what is the quality of information they find on popular AI platforms like ChatGPT or Google Bard? Are people going to AI tools for help with housing, family, debt and other problems? And do these tools give them helpful, correct, actionable information -- or are there hallucinations, mistakes, or other harms? This session will present the results of recent research on public use of AI for legal help, as well as an evaluation of AI platforms' main opportunities and risks around access to justice. It will discuss trends among various demographic groups, around whether and how they would use AI for legal help. It will also present next steps that legal services, pro bono, and court groups might take to harness the opportunities of AI while also mitigating its risks.

• Participants will learn how various community members are using AI to seek help, information, and services around legal problems they're experiencing. This can help them understand their client audience better -- and how they might adapt services for people who have tried to DIY legal research, authoring of demand letters, prep for court, and other legal tasks through AI assistance.
• Participants will learn how common hallucinations, mistakes, and other quality problems are on different AI platforms, when people ask for legal help information, services, forms, and other access to justice topics. This will help them have a more accurate view of AI platform's risks and harms. They may use this to issue client and professional guidance on whether and how to use AI -- or how to improve critical thinking and fact-checking of AI.
• Participants will discuss and learn how AI platforms might be improved regarding the legal help domain. This will include exploration of regulation, policy work, collaborations, and other ways that they may improve the quality of legal help information that people receive when they use these AI platforms.