A somewhat tepid response, the CMA did say that it overall, supported the government in its AI ambitions, and at this stage, agrees with their non-statutory approach to regulations.

However, the CMA made several nods to further conversations down the line surrounding specific regulations and where AI currently fits in their remit.

The CMA listed their intentions for regulations regarding the different sections provided in the initial whitepaper.

For consumer safety, they plan to use their current role to protect competition in the AI market to empower consumers. For instances where consumers are not able to access the technical function or security of an AI product, the CMA may need to intervene to make these clear.

The CMA also gave special notice to ensure an appropriate level of transparency and explainability.

This involves making sure AI products are clear about their database and if they are producing ‘objective’ results or results based on payment.

However, the explainability element of the CMA’s work could run into problems when faced with generative AI systems, which are not directly auditable and can produce different results to the same question without any specific reason.

For principles of fairness, the CMA believes in a context-specific definition, as fairness should consider “the context surrounding the AI system and not just the algorithm itself: for example, data collection, testing and evaluation practices.”

Essentially, the CMA may have to take into consideration fairness in AI surrounding more than purely market fairness, as it could enhance discrimination and bias rather than mitigate it.

In situations of accountability and governance, the CMA says it “holds legal persons responsible for the effects of AI systems that they deploy in relation to our remit.”

However, there may be some changes to regarding accountability for certain systems, particularly tacit algorithmic collusion – the CMA welcomes further discussion on this.

In contestability and redress, the CMA notes that “the opacity of algorithmic systems and the lack of operational transparency make it hard for consumers and customers effectively to discipline firms.”

To account for this, regulators must be well equipped to act where necessary, the CMA says.

AI Regulations 

The UK is taking its starting steps to regulate AI, with other countries and organisations attempting to do the same.

Prime Minister Rishi Sunak is set to meet with US President Joe Biden, and is set to discuss the need for AI regulations.

Recommended

• The Most Expensive NFTs Ever Sold
• UK Gov Commits £42.5m to Sustainable Agriculture Projects
• 67% of Scots SMEs Report Skill Shortages, Especially Tech Skills

PM Rishi Sunak commented: “AI clearly can bring massive benefits to the economy and society, but we need to make sure this is done in a way that is safe and secure.”

“I think the UK can play a leadership role, because ultimately, we’re only going to grapple with this problem and solve it if we work together not just with the companies, but with countries around the world.”

Following the UK’s AI Whitepaper, scientists and notable names in tech issued a letter of concern, urging developers to put the breaks on AI until regulation had time to catch up.

The EU and US are working together currently to establish a voluntary code of AI ethics they can publish soon as a guide to developers and regulatory bodies.

While the CMA is making headway on educating their regulators on AI technology, the ever evolving tech requires ongoing review.

The CMA recognised in the response that adding specific layers to AI regulation, such as new discriminatory boards, may delay regulations further, potentially causing them to fall behind as the tech continues to develop.