Redo You

As AI continues its exponential evolution, a looming question unsettles scientists, policymakers, and tech executives alike: what happens when machines surpass human intelligence in all domains? This scenario, known as Artificial Superintelligence (ASI), no longer sits in the distant realm of sci-fi. According to leading researchers, it could arrive within the next decade—and with it, an array of societal risks we’re woefully unprepared to handle.

From Buzzword to Threat: Defining ASI

Artificial Superintelligence differs fundamentally from current AI systems. While today’s AI tools—like ChatGPT, Gemini or Claude—excel in narrow tasks, ASI refers to systems with generalised intelligence that outperforms the smartest humans in logic, creativity, reasoning, persuasion, and even emotional intelligence.

Such capabilities could bring immense benefits—accelerated scientific discovery, cures for disease, climate solutions—but without adequate controls, they also pose existential risks. These range from unintentional harm due to misaligned goals, to catastrophic misuse by state or non-state actors.

The Cold War Analogy: Two Races, One Risk

As highlighted by commentators like Ilya Sutskever (former OpenAI chief scientist), the race toward ASI mirrors the nuclear arms race of the 20th century. However, it’s unfolding faster and with far fewer rules. Today’s dominant players—primarily the United States and China—are locked in two parallel competitions: one for commercial AI supremacy, and another, more quietly, for cognitive supremacy.

What makes this race dangerous isn’t just the speed—it’s the opacity. Unlike nuclear weapons, which require complex infrastructure and leave visible signatures, ASI can be built in the shadows. As a result, the traditional tools of international security—treaties, inspections, mutual deterrence—struggle to apply.

“We need enforceable agreements with verification measures—like Cold War-era arms control—to avoid disaster,” warns an op-ed in the New York Post【source】.

Policymakers Wake Up—But Slowly

In response to growing concern, some governments have begun laying the groundwork for regulation. The European Union passed the AI Act, which categorises AI tools based on risk and bans certain uses outright. But it stops short of specifically addressing ASI.

The UK is considering a more focused approach. Think tank Policy Exchange has proposed establishing a Superintelligence Council and an Office for Superintelligence to manage the unique societal impacts that ASI will unleash. The idea is to embed ASI considerations into mainstream policymaking, much like environmental or economic impact assessments.

“Governments can’t afford to play catch-up when it comes to superintelligence,” argues Ed de Minckwitz, a senior fellow at the institute. “It will reshape everything from education and employment to national defence.”【The Times】

The Private Sector: From Acceleration to Self-Regulation

Interestingly, some of the loudest alarms are coming from inside the AI industry itself.

Meta recently announced the creation of a new lab dedicated to ASI, signalling a shift in tone after earlier setbacks in their general-purpose AI efforts. Meanwhile, Ilya Sutskever has launched Safe Superintelligence Inc., a company whose singular goal is to ensure that superintelligence, when it emerges, is both safe and controllable.

This growing movement toward AI safety is a clear sign that even those building the technology fear what could happen without sufficient safeguards.

Ethics and Inclusion: Who Gets to Define ‘Safe’?

Another challenge is deciding what “safe” even means—and who decides. Western tech firms and policymakers dominate today’s discourse, but AI is a global phenomenon. Civil society advocates warn that frameworks built in Silicon Valley or Brussels may not reflect the values of Africa, South America, or Southeast Asia.

This underscores the importance of inclusive governance—ensuring that diverse voices, especially from the Global South, help shape what a safe ASI future looks like.

The Stakes Couldn’t Be Higher

At this point, it’s not just about technology—it’s about human survival and flourishing. ASI could mark the next leap in human progress or trigger a disaster unlike any the world has seen. The outcome will depend not just on how we build AI, but on how we govern it.

The clock is ticking.