The Economist’s Editor-in-Chief, Zanny Minton Beddoes, recently convened a conversation with two of the world’s foremost leaders in artificial intelligence: Sir Demis Hassabis, co-founder and CEO of Google DeepMind, and Dario Amodei, co-founder and CEO of Anthropic. In a candid exchange covering everything from near-term breakthroughs to existential global risks, they explored whether advanced AI could soon rival—or even surpass—human intellect.
At the heart of the conversation was the elusive goal of AGI (Artificial General Intelligence)—a system capable of the full range of human cognitive tasks. While both Hassabis and Amodei share a vision that this technology is on its way, they differ slightly on timelines:
Despite their differences on precisely what AGI is and when it may arise, both agreed these transformative systems will be unlike any technology we have seen.
A key theme was the global race to develop cutting-edge AI. Amodei voiced concerns that an authoritarian regime achieving AGI first could “upend the balance of power.” Hassabis echoed this unease, emphasizing the need for international discussion and cooperation—what he has previously called a “CERN for AGI,” referencing the collaborative spirit of the particle physics organization.
However, creating a robust framework for AI safety and governance remains a daunting challenge, particularly amidst simmering geopolitical tensions. The conversation highlighted the tension between accelerating research and constraining it to ensure safety and ethical development.
Both CEOs championed the development of global regulatory bodies to help monitor and guide progress in AI. Hassabis suggested that, given the technology’s transformative potential, a group akin to the International Atomic Energy Agency (IAEA) may be necessary: an institution empowered to track development and enforce safety standards.
Yet, the current international climate makes consensus difficult. While Europe leans toward heavier regulation and the U.S. pursues swift innovation, the ideal middle ground, in Hassabis’s view, involves embracing the opportunities that AI offers—especially in fields like medicine and climate science—while designing regulations to prevent misuse and mitigate risks.
Editor-in-Chief Beddoes raised a personal question for both leaders: Do you ever fear becoming modern-day versions of J. Robert Oppenheimer? Oppenheimer, dubbed the “father of the atomic bomb,” famously wrestled with his moral responsibility after unleashing the nuclear age.
Both founders strongly believe these challenges cannot be left solely to private companies. They see establishing new institutions as crucial to help oversee safety, ethics, and the technology’s direction—a system that would take the burden off a small handful of executives and researchers.
As the discussion turned to ensuring the public and governments take potential AI risks seriously, Amodei noted his team’s work testing their models in the lab for dangerous capabilities—such as the generation of bioweapon instructions or autonomous decision-making that breaks from human-aligned goals.
Should advanced AI inadvertently cause a serious crisis, the pressure to impose regulatory guardrails would become overwhelming. But their shared hope is to demonstrate hazards in controlled research settings, before a real-world catastrophe forces a hurried—and potentially draconian—response.
The conversation concluded on a more hopeful note. Both CEOs remain deeply excited about AI’s potential for scientific discovery, medical breakthroughs, and solving large-scale challenges like climate change. Hassabis envisions AI-led research unraveling new cures within the next decade. Amodei anticipates rapid gains in productivity—from writing code to autonomously running complex tasks.
In the coming year, they predict the rise of “agent-based” AI—systems that can act on instructions, learn from feedback, and achieve goals semi-independently. This shift could spark new ways to tackle time-intensive or intricate problems across industries.
This lively panel made one thing clear: AI is poised to transform the world—potentially more radically and more rapidly than any previous technology. From the race to reach AGI first, to the need for global regulation, to the personal moral questions haunting AI’s leading voices, the stakes could not be higher. Yet if guided responsibly, advanced AI might usher in an era of unprecedented innovation, extending human capability in ways once confined to science fiction.
How we navigate these simultaneous opportunities and perils will shape not just the evolution of AI, but the future of humanity. And as Hassabis and Amodei humbly acknowledged, the decisions they make—and the structures the world puts into place—will be pivotal in determining whether this powerful technology becomes a force for human flourishing or a source of existential risk.
