- Dario Amodei warns of rapid AI risks and urges urgent action by 2026.
- Demis Hassabis estimates a 50% chance of AGI by 2030 citing current gaps.
- Both agree AI's future impact will be profound and requires careful governance.
In the world of artificial intelligence (AI), the tone of public leadership is shifting. At the heart of the current debate are two of the industry's most influential figures: Dario Amodei, co-founder and CEO of Anthropic, and Demis Hassabis, chief executive of Google DeepMind. While both acknowledge AI's profound impact, their visions for the technology's future diverge in emphasis and urgency. Hassabis, a Nobel laureate, has recently struck a more measured tone on artificial intelligence at the World Economic Forum in Davos.
Amodei has issued a stark warning about the rapid advance of artificial intelligence, arguing that powerful AI systems could pose profound risks to societies and economies. In a long, reflective essay titled 'The Adolescence of Technology', Amodei described a future in which AI could soon outperform the brightest human minds across disciplines and act with a degree of autonomy that current institutions are ill-prepared to manage. He cautioned that civilisation is "considerably closer to real danger in 2026 than in 2023," urging policymakers, technologists and the public to confront these challenges urgently.
At Davos, Amodei suggested that Artificial General Intelligence (AGI) could arrive within the next one to two years, driven by AI systems that can increasingly write and improve their own code. He warned that the risks include autonomous AI systems acting against human interests, misuse in cyber warfare and bioterrorism, large-scale job losses, and the possibility of authoritarian governments using AI for mass surveillance. He also argued for strict controls on advanced AI chips, comparing their sale to hostile regimes to the spread of nuclear weapons.
Hassabis, however, struck a more cautious note. While agreeing that AI progress is accelerating, Hassabis said there is still a significant gap between current systems and true AGI. He estimated a 50 percent chance of achieving AGI by 2030, pointing out that areas such as scientific creativity, reasoning, and real-world experimentation remain difficult for AI to master.
Hassabis acknowledged that AI could disrupt entry-level white-collar jobs but emphasized the need for international cooperation among governments, economists, and scientists to manage the transition responsibly.
Despite their differences, both leaders agreed that AI will have profound economic and geopolitical consequences and that decisions made in the next few years will shape how safely the technology is deployed.
What is Artificial General Intelligence (AGI)?
AGI takes AI a step further. While AI is task-specific, AGI aims to possess intelligence that can be applied across a wide range of tasks, similar to human intelligence.
In essence, AGI would be a machine with the ability to understand, learn, and apply knowledge in diverse domains, much like a human being.
