Thomas Unise
A well-intentioned AI chatbot designed to simplify probate procedures for Alaska’s court system has proven to be a significant disappointment during testing, plagued by hallucinations and user frustration.
The Troubled Development of Alaska Virtual Assistant
The Alaska Virtual Assistant (AVA) was developed to help people navigate the complex probate process following a loved one’s death. However, what was meant to be a three-month project has been stuck in development for over a year due to persistent issues with the AI’s functionality.
The project, a collaboration between the Alaska Court System and LawDroid (an AI legal tools company), encountered multiple challenges that highlight the current limitations of AI in sensitive legal applications:
- The chatbot frequently hallucinated information, providing false facts despite being restricted to its knowledge base
- It incorrectly claimed Alaska had a law school and suggested users contact its alumni network
- Users found the bot’s artificial expressions of empathy and condolences irritating rather than helpful
- The development process proved far more labor-intensive than anticipated
Scaled-Back Expectations
After extensive testing and refinement, project leaders have significantly lowered their expectations for what the AI can actually accomplish:
“We wanted to replicate what our human facilitators at the self-help center are able to share with people. But we’re not confident that the bots can work in that fashion, because of the issues with some inaccuracies and some incompleteness,” explained Stacey Marz, administrative director of the Alaska Court System.
The team has made specific adjustments based on user feedback, including removing condolences that test users found excessive and unwelcome from an AI system. Despite the challenges, AVA is finally scheduled for public launch in late January after substantial revisions.
Reality Check on AI Promises
This case serves as a sobering counterpoint to the widespread enthusiasm about generative AI’s potential to revolutionize legal services. Marz noted the disconnect between the hype around AI “democratizing access to the courts” and the difficult reality of implementing reliable AI systems for sensitive legal matters.
The AVA project illustrates a common pattern with emerging AI technologies: initial promises often exceed what the technology can reliably deliver, particularly in specialized domains where accuracy is paramount.
Conclusion
The Alaska Virtual Assistant’s troubled development highlights the significant gap between AI’s theoretical potential and its practical implementation in specialized fields like law. While the technology continues to advance, this case demonstrates that AI systems may require more extensive development and testing than anticipated before they can reliably serve in sensitive roles where accuracy and appropriateness are essential.
GIPHY App Key not set. Please check settings