A.I. Series: Shaping Regulation for Artificial Intelligence
According to the US International Trade Administration, the UK AI market is worth more than £16.8bn and is expected to grow to nearly £800bn by 2035. Despite these strong economic projections for growth, the public still have several concerns surrounding AI, with ethical implications and data security among their top concerns. AI systems handle vast amounts of data, raising concerns about privacy violations and potential misuse of personal information. Following its consultation from the AI white paper, the UK Government has adopted a cross-sector and outcome-based framework for regulating AI, underpinned by five core principles. These principles have been adopted by existing regulators such as the Information Commissioner's Office (ICO), Ofcom, and the FCA as they supervise AI within their respective domains. With technology advancing quickly how can the Government ensure that regulation of AI is robust whilst also allowing for a pro-innovation led AI strategy?
Key Discussion Points:
- We regulate the use of AI systems but not the development – if we only listen to the people who develop it there will be biases present.
- Regulation is a socio-technical problem. How will A.I. be used and what effects will these uses have? We have seen potential for misinformation and disinformation through deepfakes in the 2024 General Election, and it can be taken further into people’s lives through crimes as awful as revenge porn.
- International comparisons. In America there is a scheme that places people with PhD understanding of AI in the room with lawmakers, allowing them to “speak the same language” when discussing key issues. In the UK there is a clear gap between the pace of AI machine learning and innovation that legislators can’t keep up with
- The importance of transparency and collaboration. Different regulators have different levels of knowledge, different sectors have different approaches, different businesses have different principles and different people have different opinions.
- America’s AI regulation is very flimsy as it relies on an executive order which can easily be overturned by an incoming president. Meanwhile the EU Artificial Intelligence Act has many shortcomings, including a lack of principle and agility, and took 5 years to be enacted into law.
- How best Governments can procure AI systems? There is a need to look at the processes and relationships between suppliers and public sector procurers to look to improve specifications and achieve best and safe outcomes from AI.