On July 18, 2024, the Association of eDiscovery Specialists (ACEDS) hosted an informative webinar titled “Transforming eDiscovery and Investigations with AI: Addressing Cost, Scale, and Transparency.” Moderated by ACEDS President Mike Quartararo, the event featured industry experts from Hanzo and IBM who shared insights on the future of artificial intelligence in legal technology. Following are some highlights.

AI in Legal Tech: Setting the Stage

Mike Quartararo kicked off the webinar by providing an overview of the topic and introducing the panelists. “Our goal is to talk about the future of eDiscovery and investigations, as powered by AI,” Quartararo remarked. He emphasized the focus on secure, transparent, and practical AI solutions.

IBM’s Perspective: Trust and Security

Omkar Nimbalkar, IBM’s Vice President for Ecosystem Engineering, highlighted the collaboration between IBM and Hanzo, underscoring the importance of trust and security. He explained that trustworthy AI ensures accuracy, reliability, transparency, and ethical considerations, which are foundational for successful AI implementation in enterprise settings.

Nimbalkar highlighted IBM’s Granite model, which integrates machine learning and generative AI capabilities on a single platform emphasizing that the Granite model is developed on IBM’s OpenShift and Red Hat technologies, providing enterprise-grade security and scalability, making it suitable for a wide range of applications. By filtering datasets to remove duplicates, hate speech, and profanity, the Granite model upholds high standards of ethical AI. Nimbalkar concluded, “IBM’s Granite model merges the power of machine learning and generative AI in a single platform. We strongly believe in trust, and therefore, Watson is truly the best platform for AI in business.” This blend of advanced capabilities and rigorous security measures demonstrates IBM’s leadership in delivering dependable and reliable AI solutions.

Hanzo’s Approach: Tackling Hallucinations and Data Complexities

Aidan Randle-Conde, Director of Data Science at Hanzo, provided a detailed explanation of the challenges posed by AI hallucinations and the complexities of handling conversational data in eDiscovery. He described hallucinations as instances where large language models (LLMs) generate misleading or irrelevant text due to ambiguous or open-ended input prompts. “If you allow the LLM to keep generating text, like I’m generating words right now, the longer that goes on, the more likely it is to hallucinate,” Randle-Conde explained. To mitigate these risks, Hanzo focuses on crafting precise and specific prompts and implementing governance mechanisms to ensure accurate AI outputs.

Revolutionizing eDiscovery: Practical Applications of AI

Dave Ruel, VP of Product at Hanzo, highlighted that Spotlight AI can efficiently process millions of messages, significantly reducing the time and effort required for eDiscovery. “Our benchmarks are showing sixteen times improvement versus a human review,” Ruel stated. By integrating AI with traditional (hybrid) methods like keyword searching and facet searching, Spotlight AI elevates the most relevant data while minimizing noise. This hybrid approach ensures that legal professionals can quickly identify critical information, enhancing both the speed and accuracy of the eDiscovery process. Ruel added. “We can control costs on our end really, really well by using combinations of LLMs and small language models (SLMs).”

Future Directions: Scalability, Trust, and Cost Efficiency

Looking ahead, the panelists shared their visions for AI in legal tech. Randle-Conde predicted a future where data volume becomes less of an issue. “We’ll be able to process arbitrarily large datasets, and there is no upper limit on what can be reviewed by humans assisted by LLMs in a fixed period of time,” he said. Nimbalkar stressed the significance of trustworthy AI, adding, “IBM’s governance platform offers an AI fact sheet that captures various metrics for quality, accuracy, model health, and compliance.” Ruel commented on the trend of AI getting more cost-efficient saying, “I’m gonna go down the road of, bigger, better, faster, cheaper.”

Closing Remarks: Encouragement for AI Adoption

As the webinar drew to a close, Ruel reassured attendees about their position in the AI adoption curve. He cited a statistic from the IBM Think conference, highlighting that over fifty percent of enterprises are still in the proof-of-concept or testing phase with generative AI. “Don’t panic. Don’t feel like you’re behind,” Ruel advised. “This is a journey we’re all on.”

Conclusion

The ACEDS webinar provided a comprehensive look at how AI is set to transform eDiscovery and investigations. Panelists from Hanzo and IBM highlighted the importance of trust, security, and cost-efficiency while exploring the advancements and future possibilities of AI technology. As the legal industry embraces AI solutions, these discussions pave the way for more effective and efficient eDiscovery practices.

[View source.]