It doesn’t seem that long ago that Cyber-security, applications or Cloud technology were the main topics of conversation for law firms outside of legal matters. But now the topic that crops up most, is artificial intelligence [AI] in all its guises, the helpful, unhelpful, benign and malignant.
Questions are being asked about the use of AI in law firms and a range of issues are bubbling to the surface. Many are suggesting that AI suffers from an accuracy and bias issue, which can cause AI to offer incorrect and possibly harmful results depending on the way questions are framed.
When it comes to incorrect or harmful outcomes created through us of AI, it’s important solicitors remember that ultimately, they are still accountable to clients for the services provided, whether external AI was used and the client is aware or not.
The EU acts and the UK will follow
In June, the EU overwhelmingly voted to adopt the proposed Artificial Intelligence Act (AI Act), which aims to regulate the use and development of artificial intelligence throughout the EU. Discussions are now to take place between the EU and the UK to agree the text of the AI Act, with a view to this becoming law by the end of the year.
Whilst the EU AI Act adopts a different approach to the UK’s current proposals, set out in the government’s white paper in March 2023, both are designed to help deliver the world-leading regulation that will drive innovation and help us lead the world in AI.
Even though the UK and EU have taken different approaches, overall both are seeking to balance the risks to users and the general public, whilst supporting investment and innovation. However, the UK’s white paper does not highlight plans to introduce legislation to cover AI concerns specifically.
The UK white paper proposes five principles of AI governance, which in no particular order, include:
Safety, security and robustness – AI systems must operate in a robust, secure and safe manner throughout their lifecycle.
Appropriate transparency and ‘explainability’ – AI systems must provide transparency, which refers to making appropriate information about the AI system available to the relevant individuals. It must also provide ‘explainability’, which refers to how relevant parties can access, interpret and understand the decision-making processes of an AI system.
Fairness – AI systems should not undermine the legal rights of individuals or organisations, discriminate unfairly against individuals, or create an unfair market.
Accountability and governance – Governance measures should ensure effective oversight of the supply and use of AI systems, with clear accountability established across the AI lifecycle.
Contestability and redress – Anyone impacted by an AI system, should be able to contest an AI decision or outcome they believe is harmful or likely to create a risk of harm.
The White Paper suggests there are at least two audiences that an AI developer must consider, firstly those who may be affected by the AI system and the regulator who requires sufficient information to perform its regulatory oversight effectively.
The UK white paper also does not propose a specific approach to liability, but does recognise it is an important consideration and will consult experts, technicians and lawyers to consider how existing frameworks can be adapted for application to AI systems.
AI training causes copyright concern
The UK’s white paper has stirred controversy in relation to copyright, believing existing intellectual property law would deal with copyright protection for computer-generated works. There was also a proposal to exempt text and data mining, which would remove copyright holders’ rights to charge licence fees for commercial usage and is opposed by the creative and academic industries.
However, now Government is pledging to work closely with AI developers and rightsholders to produce a code of practice that covers access to and use of copyrighted works train AI systems. This raises the potential for rightsholders to grant licences to AI developers wanting to use such materials to train their AI systems.
Considerations for business
The proposals for regulating AI in the UK are still being developed, but the pace of advance of the AI industry is easily outstripping the speed at which the proposed regulations can be drafted and amended in light of these advances.
This might be one reason the UK white paper is slightly vague, some may say ambiguous, in its proposed regulations, to ensure the environment is not too restrictive and allow the industry to shape AI regulation in the UK, which may also influence the EU regulations in the future.
Businesses developing AI solutions in the UK or the EU must consider how they will deliver robust policies and frameworks, capable of balancing innovation to provide valuable services for their customers, with compliance to satisfy the needs of the relevant jurisdiction’s regulators.
Major considerations for organisations expecting to deploy AI solutions, include identifying what AI solutions are being used and what are their intended purposes. It is also important to understand how the AI solutions have been trained and if obvious gaps in training data could create system bias.
Organisations operating in regulated sectors should use existing compliance frameworks to document appropriate risk assessments, which should satisfy the UK white paper principles, given that the overall objectives are similar.
The approach to compliance with the UK white paper AI principles and regulation, must be considered with other factors such as data protection compliance and a review of intellectual property and liability issues relevant to use of AI systems.
Where do we go next?
The possibilities with AI appear endless and while the initial excitement and novelty value has perhaps already worn off, there are already large tech companies pursuing small specialist AI businesses as the future capability, adaptability and profitability becomes clearer.
The productivity benefits of artificial intelligence may come to all our workplaces soon, which will undoubtedly benefit a large proportion of the working population, if it allows them to do the same amount of work they do now, but in less time. Maybe the four/three day working week is possible.
Even while researching and writing this, AI will have advanced further and faster than the authorities can when developing a framework to regulate it, so for now, watch the space with interest and speak to a lawyer who understands your challenges on a personal level.
