Artificial intelligence (AI) is more than just technology; it’s a socio-technical system that could exacerbate social inequalities and consolidate power in the hands of a few.
At the recent Digital Ethics Summit hosted by TechUK, experts pointed to 2025 as a pivotal year for AI, even amid a lull in groundbreaking technical advancements since the surge of generative AI in late 2022. Companies now dive into the practical aspects of integrating AI into their operations. The conversation highlighted a shift toward deploying these applications effectively and scaling their use while prioritizing safety.
The new UK Labour government aims to harness AI for economic growth and promote greater diffusion across sectors. However, the rapid introduction of AI tools we saw in 2024 brings an urgent need for concrete, standardized ethical frameworks. Existing guidelines are often too vague and don’t address the socio-technical nature of AI, which intertwines social dynamics and technical capabilities. Trust in both government and corporations is dwindling, adding urgency to the dialogue about how to develop AI responsibly.
In this landscape, inclusivity and participation in AI development are crucial. Many delegates emphasized that without diverse voices in the conversation, we risk deepening existing inequalities. They noted that only China and the U.S. can make unilateral decisions about AI development, raising questions about whether an open or closed approach to technology is more advantageous.
In a panel discussion about applying AI ethics in real-world settings, KPMG’s Leanne Allen pointed out that while ethical principles exist, implementing them remains complicated. For instance, explaining outputs from generative AI is challenging, signaling the need for more nuanced guidelines. Melissa Heikkilä of MIT Technology Review echoed this sentiment, stressing the lack of consensus on auditing AI effectively, which hinders standardization.
Alice Schoenauer Sebag from Cohere highlighted initiatives like MLCommons to create a shared understanding and benchmarking for AI safety. As organizations shift from experimenting with AI systems to deploying them, conversations are becoming more targeted and specific.
Allen noted that most companies rely on existing AI solutions rather than creating their own, leading to a demand for reassurance on their safety and reliability. However, significant cultural differences impact the focus on English-speaking contexts in the industry. Sebag urged for a broader view of safety that accommodates diverse meanings across cultures.
Heikkilä warned against the risks of consolidating power further among a few firms and nations. The conversation surrounding governance is complicated by geopolitical tensions, especially between the U.S. and China. Alex Krasodomski from Chatham House stressed the necessity for governments to become AI builders, not just regulators, to level the playing field.
Andrew Pakes, a Labour MP, pointed out the internal disparities within the UK related to AI and how this technological wave might impact people’s lives. He expressed concern that the rapid changes in industrial practices could lead to greater societal division if not managed inclusively.
The need for collaboration when deploying AI in the public sector resonated among participants at the summit. Jeni Tennison suggested engaging people actively in the AI development process to ensure that their values and needs are integrated from the start.
Looking globally, Martin Tisné emphasized the importance of international cooperation to address the strategic significance of AI. Countries need to see AI not just as a competition but as an opportunity for collaboration.
Krasodomski also pointed out that strengthening national capabilities in AI could empower governments to engage with major tech firms more effectively. Countries like Sweden and Switzerland are already investing in their own AI models.
Chloe MacEwen from Microsoft highlighted the UK’s strong foundation for AI innovation through its scientific talent and commitment to cloud infrastructure. However, Linda Griffin from Mozilla cautioned against over-reliance on a few dominant companies for AI’s infrastructure, which she sees as a significant issue.
In discussing open versus closed AI models, Griffin noted the need to keep AI development accessible and collaborative, drawing on lessons from the fight to maintain an open internet. She advocated for transparency in AI practices to build public trust, especially in sectors like healthcare.
Overall, the conversations reflect a growing awareness of AI’s potential impacts, the need for ethical clarity, and the importance of inclusive engagement across different cultures and communities.
