News You Can Use

Annual Law Department Operations Survey. A report on a survey from The Blickstein Group and FI covering the effect of AI on legal service delivery. Showed an increase in legal ops roles in in-house teams. Also indicated a lack of maturity on AI strategy, as well as continued growing pains with CLM. [Only 83 participants]. Broadly positive and optimistic outlook as to the effect of AI on legal teams, but conscious that they are not well prepared. Primary driver of AI strategies was efficiency and performance (81%) and cost savings (42%). 37% of respondents thought their law firms were innovative, whilst 43% thought that their firms were leveraging technology properly. 65% thought that GenAI adoption would lower their law firm spend and 87% thought that it would help them bring more work in-house. 89% thought that in-house teams would be the primary driver of innovation and change in the legal sector!

Despite only 83 participants, this survey is a good indicator of general sentiment. It shows that there is a huge amount of opportunity as seen by clients to drive forward new pricing mechanisms and adopt technology as a law firm, with a really positive outlook on the effect of AI. It also showed a clear desire to do more work in-house, with the reasons for external counsel mainly being bandwidth related. The belief that AI will let clients in-source more work is something to be aware of. The lack of sophistication in AI strategies is a great sell for our workshops on AI, but also the desire to change how law firms work with clients shown here could be something to push on from MLS.

Open AI's o3 introduces a new level of reasoning, decision-making, and task execution, achieving human-equivalent scores in areas like math, logic, and creative problem-solving. Its performance shows significant strides in handling complex, multi-step processes. However, the model introduces significant questions around ethical use and risk, particularly in high-stakes industries. Key concerns include autonomy risks, bias amplification, regulation and accountability, and security risks.

OpenAI o3's advancements could support AGPT in more complex legal analysis, such as identifying nuanced contract risks or creating predictive models for legal outcomes. It may be that this could be a "supercharged AGPT" version due to high costs and suitability to complex issues. However, we would need to implement robust guidelines, such as explainability, traceability of AI decisions, and mechanisms to audit AI outputs before deployment to ensure client trust and regulatory compliance.

The report highlights the importance of balancing capability with safety in AI agent design. It emphasises training methods that encourage agents to remain collaborative and aligned with human goals, even when handling complex or ambiguous tasks. The research also includes techniques such as hallucination reduction frameworks, reinforcement learning with human feedback (RLHF), and controlled exploration methods to mitigate risks like power-seeking behaviours or hallucination errors in AI agents.

These techniques can directly guide AG's development of safe and reliable AI agents for AGPT, particularly in areas like contract analysis or deal management. For example, RLHF could improve our AI's ability to align its outputs with the expectations of legal professionals. A key risk with Agents is that one mistake snowballs into many, so understanding the infra and how to build them will be important in ensuring reliability.

The article explores practical applications of AI in the legal sector, including tools for contract drafting, legal research, and compliance management. It also discusses the potential for AI to enable more personalised and scalable client services. However, it notes that adoption rates vary, with smaller firms (25%) often lagging behind due to cost and complexity concerns.

This aligns with AG's mission to leverage AI for both operational efficiency and client-facing innovation. By showcasing our tools as accessible and scalable, we can appeal to a wider range of clients, including those who may be hesitant about AI adoption. Additionally, these statistics reinforce the importance of demonstrating tangible ROI and time savings during client pitches to build confidence in AI's value.