Local authority legal departments and AI

Thanh Lanh-Connolly looks at how local authority legal teams can harness AI power to improve efficiency in commercial and corporate matters.

Local authorities are under an immense pressure to reduce costs whilst increasing productivity and efficiency to serve the public, and legal departments within those local authorities are no exception. Local authority lawyers have many responsibilities to fulfil, e.g. ensuring that they meet their wider duties to the public as well as providing internal support in respect of corporate and operational matters. Lawyers may find themselves concurrently dealing with a wide range of commercial contracts and corporate documents (such as governance documents, policies, frameworks, or constitutions of companies they own). The volume and complexity of the work could be significant, especially against the backdrop of challenging timescales and budget and resource constraints.

This article explores how AI tools may help to ease these pain points, and the potential pitfalls to be aware of and managed when adopting these tools.

A powerful tool

Preparing commercial contracts and corporate documents is a long and arduous process. It takes time to put the first draft together, which is no where near the final product. Back-and-forth correspondence and negotiation with relevant stakeholders are a given before the documents can be finalised. AI tools can help to cut down the time and resources spent on this process. Some examples are:

• AI tools can help extract and analyse historical data on the whole lifecycle of past documents, combining with current market trends and future demand forecasts to identify optimal requirements and specifications for new documents.
• Where standardised documentation is used, the system can generate the first drafts with necessary customisation on the basis of input variables with minimal human intervention.
• An AI-powered workflow and contract management system can help automate the approval and authorisation process, validate information and flag anomalies, both internally right through to the contract signing stage.

AI can perform the above tasks with high accuracy and in a matter of minutes. This will free up local authorities’ human resources to spend on other pressing priorities, reducing backlogs and improving the quality of the work environment and wellbeing of its members of staff. This will in turn increase the authorities’ productivity and efficiency in serving citizens.

To use with caution 

AI systems rely on a complex ecosystem, including appropriate digital infrastructure and a large volume of reliable and quality data. This could mean high implementation costs, not only in terms of the cost of the relevant solution, but also the amount of time and resource required to set up an adequate environment for the system to function properly. Such set up may entail integrating with or replacing the legacy systems, digitalising paper-based data, re-organising, cleansing and standardising historical data as well as training members of staff to manage the transition and use the new systems. Issues such as confidentiality, data protection, intellectual property, exit plan etc. will need to be dealt with at the outset and throughout the life cycle of AI.

Currently language-based automation tools (including generative AI) cannot understand the context and meanings of words. While they can produce plausible strings of text on the basis of processing a large quantity of text, there is no guarantee that the outputs are correct. Errors or problematic results may come from both the underlying dataset and the algorithm of the systems. For example, an AI tool developed on the basis of US laws may fail to provide documents relevant to legal concepts and requirements under English laws. When it comes to document review or negotiation, AI tools may be unable to look at the documentation as a whole to pick up the nuances of drafting or understand how different parts of the same document or a suite of documents work together. The government’s Generative AI Framework for HMG has covered the limitations of this technology; the principles still work for other automation tools.

Responsible use of AI tools also requires investment in the workforce. Upskilling the relevant staff with digital skills is not the only task. Local authorities would also need to manage the risk of overreliance or overestimating AI’s performance, and other human’s cognitive biases that could hinder human oversight. For example, there has been discussion that having a blind trust in an automated system over time may lead to a decrease in professional skills due to lack of initiative, frequent practice and exposure.

Conclusion

Powerful tools such as AI are usually considered a double-edged sword that needs to be handled wisely. If things go wrong, local authorities will be liable for the aftermath, not the machine. Local authorities may consider developing adequate policies and guidance on the use of these tools (citizens will not stop using publicly-available AI and AI has already been used in the public sector) and address the relevant risk. These documents would need to be crafted in a way that is relevant, suitable and implementable in the relevant organisations. This is where the power of humans comes in, to skilfully utilise these tools while providing necessary guardrails to prevent unintended loss and damage.

Thanh Lanh-Connolly is a Chartered Legal Executive in Ashfords’ Commercial team with a tech focus including AI. Ashfords Commercial team works closely with the Public Sector team to provide support to private and public bodies including local authorities on the application and implementation of the public procurement rules throughout the whole procurement lifecycle.