News You Can Use

This update combines conversational capabilities with real-time search functions. Users can query live web content and cross-check answers directly, making the tool more reliable for dynamic, up-to-date tasks like research and analysis.

The possibility of integrating similar search capabilities within AGPT could elevate its potential as a real-time legal assistant, enabling on-demand access to up-to-date laws, regulations, and case law. It would also enhance stakeholder-facing interactions by allowing instant responses backed by verified data.

The template is a structured framework to craft better prompts for AI in educational settings. It focuses on maximising AI's ability to explain complex concepts, suggest ideas, and even provide constructive feedback on student work.

This is relevant to AG's internal training initiatives. Incorporating such structured AI prompts could enhance how we train our lawyers for prompt engineering, train on new legal standards, and provide feedback on drafted contracts or reports.

The findings reveal that access to Copilot leads developers to shift their focus towards core coding activities, reducing time spent on non-core project management tasks. This shift is attributed to an increase in autonomous work and a higher engagement in exploratory activities, with the most significant benefits observed among developers of relatively lower ability. The paper suggests that generative AI has the potential to flatten organizational hierarchies by enabling workers to concentrate on creative and technical tasks, thereby enhancing productivity and innovation in knowledge-intensive industries.

The findings reaffirm the importance of having a tech stack to work alongside. Software Development is a similar 'technical expertise' based role, and the report here shows that freeing up time leads to better engagement in activities which we would class as higher value. Removing the admin or low value work from lawyers day to day should have similar effects, ensuring that teams can focus on strategic, high-value legal tasks while GenAI handles the basic work. Interestingly the evidence that lower ability developers improved more could indicate that there is more value in upskilling our trainees and juniors.

These articles argue that LLMs may have hit a plateau in their effectiveness and scalability due to issues like resource consumption, accuracy, and data limits. They call for alternative methods and frameworks to push AI innovation beyond current boundaries. Some of these include modular AI systems that focus on specialised tasks rather than general capabilities, integration of symbolic AI to enhance reasoning and interpretability, and low-resource models designed to reduce costs and energy demands.

This presents a cautionary note as we develop AG's AI tools. It highlights the importance of not relying solely on large, generalist LLMs but also exploring hybrid approaches like combining modular AI for focused legal tasks or incorporating symbolic AI to improve reasoning in niche applications.

DLA Piper is leveraging Microsoft 365 Copilot to automate routine document tasks, enhance collaboration, and reduce time spent on administrative work.

This example underscores the value of integrating productivity-focused AI like Copilot at AG, ensuring we stay competitive while maintaining efficiency in handling legal matters. This is worth being aware of in light of the continued push by Clifford Chance recently, some firms are going "all in" with Microsoft Copilot.