Alaska’s court system is grappling with the challenges of implementing a generative AI chatbot designed to assist residents with probate procedures, a process already fraught with emotional complexities. Dubbed the Alaska Virtual Assistant (AVA), the initiative has been under development for over a year, significantly longer than the initial three-month timeline, primarily due to the necessity for accuracy in a high-stakes legal environment.
Aubrie Souza, a consultant at the National Center for State Courts (NCSC), emphasized the challenges in getting AVA right, stating, “We are now at well over a year and three months, but that’s all because of the due diligence that was required.” The AVA project aims to simplify the probate process, which involves transferring property from a deceased individual, with the goal of making legal assistance more accessible.
Stacey Marz, the administrative director of the Alaska Court System, expressed the need for complete reliability in AVA’s responses. “With a project like this, we need to be 100% accurate, and that’s really difficult with this technology,” she said. The stakes are high; any misinformation could have severe consequences for families navigating the complexities of probate.
The experience of developing AVA highlights broader issues that government agencies face when adopting AI technologies. A recent Deloitte report revealed that less than 6% of local government practitioners prioritize AI in service delivery, reflecting widespread hesitancy rooted in concerns about reliability and the role of human oversight. Marz noted that while AI could offer efficiencies, “we’re not confident that the bots can work in that fashion” when compared to human facilitators who provide support at self-help centers.
The vision for AVA was inspired by Alaska’s family law helpline, which offers free guidance on issues ranging from divorce to protective orders. Marz and her team aimed to replicate this experience through AI, creating a self-help tool that would guide users through legal processes. The project received initial funding from NCSC, with Tom Martin, a lawyer and law professor, leading the technical development through his company, LawDroid.
Martin underscored the importance of crafting AVA’s personality to ensure it remains helpful and straightforward. He remarked, “For a legal application, you don’t want that. You want it to be rule-following but smart and able to explain itself in plain language.” Early iterations of AVA faced challenges, including the phenomenon of “hallucinations,” where AI confidently produces false information. For example, AVA inaccurately referenced a non-existent law school in Alaska as a resource for legal help.
To mitigate such issues, Martin has implemented strict parameters, confining AVA’s knowledge to the Alaska Court System’s probate documents. This approach aims to prevent the chatbot from sourcing erroneous information online. The team has also developed a test of 91 questions to evaluate AVA’s accuracy, although time constraints led them to refine it to a set of 16 critical queries.
Cost considerations also loom large in the implementation of AVA, as tight budgets affect the feasibility of deploying advanced AI solutions. Martin pointed out that operational costs have significantly decreased; under one configuration, 20 AVA queries could cost as little as 11 cents, making it a financially viable option for the court system.
As the AVA project approaches its expected launch in late January, Marz maintains a cautious optimism about the potential benefits it could provide to Alaskans. However, she acknowledges the limitations of current AI technology and the importance of human oversight. “We wanted to replicate what our human facilitators… are able to share with people. But we’re not confident that the bots can work in that fashion,” she said, hinting at the ongoing challenges that lie ahead in balancing technological promise with the need for precision.
Despite the hurdles encountered during its development, the AVA project underscores a critical intersection of technology and public service, illustrating both the potential advancements and inherent risks in leveraging AI to enhance access to justice.
See also
India Orders X to Address Grok’s Obscene AI Content Within 72 Hours
European Banks Cut 200,000 Jobs by 2030 as AI Drives Efficiency Gains
AI Bubble Concerns and Diverging Policies Shape 2026 Asia Stocks Landscape
Musk Warns Grok Users: Illegal Content Will Face Legal Consequences Amid Indian Crackdown
California Law Requires AI Companies to Disclose Disaster Plans and Risk Assessments
