DeepMind CEO urges AI “guardrails” for responsible development

Demis Hassabis, the CEO of Google’s artificial intelligence arm, DeepMind, delivered a thought-provoking speech at London Tech Week, highlighting the immense opportunities and risks that accompany the advancement of artificial intelligence (AI). Speaking to a large audience, he stressed the need for caution and the establishment of “guardrails” in AI development.

As a prominent figure in the AI industry, Hassabis serves as a government adviser on AI and was among the experts who signed a statement emphasizing the global importance of mitigating the risks associated with AI to avoid potential extinction-level consequences comparable to nuclear war and pandemics.

During the conference, Hassabis engaged in a conversation with UK Prime Minister Rishi Sunak, discussing the challenges posed by AI. He raised concerns about the unauthorized use of AI technologies by malicious actors and the rise of deep fakes, suggesting the implementation of safeguards such as encrypted watermarking to combat these issues.

Hassabis acknowledged that there are still uncertainties surrounding AI, particularly in areas such as system alignment, controllability, and interpretability. He emphasized the necessity for thorough research to better understand these systems and their boundaries. By adopting a precautionary approach and employing methods such as the scientific method, he believes it is possible to make informed decisions and effectively manage the risks associated with increasingly powerful AI systems.

Furthermore, Hassabis emphasized the significance of international cooperation and collaboration in navigating the complex landscape of AI. He highlighted the need for discussions among countries, companies, and experts to ensure the benefits of AI are harnessed while minimizing potential risks.

To facilitate these conversations, Hassabis mentioned the upcoming global summit on AI safety, which will gather experts from around the world. The summit aims to address concerns, plan for the future, and establish the necessary guardrails and research initiatives to responsibly steer the development and deployment of AI technologies.

Prime Minister Rishi Sunak echoed Hassabis’ sentiments, emphasizing the importance of considering existential risks and implementing appropriate guardrails to ensure the safe and secure utilization of AI. He stressed the need for thorough planning, evaluation, and continuous research to effectively manage potential risks associated with AI technology.

Overall, Hassabis’ speech shed light on the critical need for responsible AI development, emphasizing the significance of global cooperation, research, and precautionary measures to harness the transformative potential of AI while safeguarding against potential risks.