To make the most of your techUK website experience, please login or register for your free account here.
As generative AI tools surge in popularity and the global conversation on AI continues to gain momentum, seamless collaboration between businesses and governments is needed to agree on the rules of the road for this fast-moving technology. What is clear is that any guidelines or policies must ensure accountability, trust, explainability, safety, and transparency.
IBM supports worldwide leaders who are pressing for action to ensure AI systems are built and used responsibly; it is a priority we share.
UK Prime Minister Rishi Sunak’s initiative to convene the first major global summit on AI safety is opportune. With the EU’s landmark AI Act entering a crucial legislative phase in Brussels, a recent push for comprehensive legislation in Washington, and governments from Canberra to São Paolo considering similar rules, it is time for a global policy conversation on AI.
The UK-hosted summit is an opportunity for governments to move forward conversations on how to govern the safe use of AI, ensuring that innovation is conducted safely and in ways people trust. Agreeing on a global set of guardrails to prevent the current and longer-term risks of AI will be challenging – but it is also necessary and possible. Lawmakers should focus on aligning around three key policy priorities:
First, the regulatory focus should be on specific high-risk use cases of AI – how the AI system is used and by whom, not on the algorithms themselves. This means creating clear, pragmatic, and effective regulations that evolve as AI does and that safeguard against ethical concerns around AI – misuse, misinformation etc. It is also about enabling the enormous positive social, economic, and environmental benefits of this revolutionary technology.
Instead of blanket regulations, policymakers should focus on regulating the use case of the AI product or service at the point of risk – where it interacts with people. An AI system that helps inform credit or employment decisions, for example, should be regulated more heavily than a chatbot that offers restaurant recommendations.
Second, there should be no room for anyone to be deceived or misled when interacting with an AI system. There needs to be a better understanding of how these systems are trained, what data they rely on, how they arrive at conclusions, and how they perform when tested for bias. To uphold principles of explainability and trust, policymakers should not grant broad exemptions from liability through licensing regimes.
Finally, as regulatory discussions evolve – here in the UK and beyond – a broad community of stakeholders should be consulted when developing AI regulation, to ensure models are shaped by diverse perspectives. Our future use of AI should be determined by the many, not the few. To that end, an open AI ecosystem promotes healthy competition, innovation, skill building, and security.
AI has the potential to revolutionise every industry sector from healthcare to finance to agriculture and beyond. While it is clear that regulating AI comes with a very real set of challenges, it is within our control. It is paramount that businesses, researchers, and civil society all work together with lawmakers on a set of guardrails for AI that protect people and promote innovation. We look forward to furthering this conversation this week in the UK.
