AI Leaders Say Human-Level Systems Are Approaching Fast

Rapid progress in artificial intelligence is forcing governments and institutions to confront a much shorter path toward human-level systems than previously expected. Industry leaders now say the gap between today’s tools and artificial general intelligence is narrowing quickly. As development accelerates, concerns around jobs, governance, and economic stability are moving to the center of the debate.

Amodei Says Human-Level AI Could Arrive Within Years, Not Decades

At the World Economic Forum in Davos, Anthropic CEO Dario Amodei warned that policymakers may not be prepared for how fast advanced AI is approaching. Appearing alongside DeepMind CEO Demis Hassabis, Amodei argued that social systems and labor markets are unlikely to adapt at the same pace as technical progress. In his view, preparation time is shrinking rather than expanding.

Amodei reiterated his belief that human-level AI is likely only years away. He said his earlier projections still stand and that progress has continued along a steep curve. Superhuman capabilities, according to his estimate, could arrive as soon as 2026 or 2027. In his words, it is difficult to see how development could stretch much beyond that timeframe.

Much of this speed comes from AI systems increasingly assisting in their own development. At Anthropic, Amodei said, software engineers are already shifting from writing code to supervising AI-generated output. Engineers now spend more time reviewing and correcting code than producing it from scratch. Within six to twelve months, he suggested, AI models may handle most coding tasks from start to finish.

Several forces are pushing this cycle forward:

• AI models now generate large portions of production-level code.
• Engineers act mainly as reviewers rather than primary authors.
• Training improvements feed directly into faster model upgrades.
• Hardware supply limits speed more than research capability.
• Shorter development cycles compress adoption timelines.

DeepMind’s Hassabis Puts AGI Odds at 50% by 2030

While acknowledging strong progress, Hassabis argued that not all fields are equally suited to automation. Areas such as coding and math are easier targets because results can be quickly verified. Other disciplines, especially the natural sciences, rely on experiments that require time and resources.

Scientific discovery, he said, remains a major barrier. Current systems can solve well-defined problems but struggle to generate new questions or theories. Producing original hypotheses, in his view, represents one of the highest levels of human creativity. AI has yet to demonstrate reliable ability in this area, and it remains unclear when—or if—that gap will close.

Because of these limits, Hassabis placed the odds of reaching AGI by 2030 at roughly fifty percent. He pointed to the difference between fast computation and genuine innovation as a key uncertainty. Even so, both executives agreed that economic disruption is no longer a distant concern.

White-collar roles are increasingly exposed. Amodei has previously estimated that up to half of entry-level professional jobs could disappear within five years, and at Davos, he stood by that figure. Office-based work, once seen as protected, now faces automation pressures similar to those that reshaped manufacturing decades earlier.

Hassabis warned that even cautious economic forecasts may underestimate the speed of change. Five to ten years, he said, is not a long time for societies to adjust. Institutions built for slower transitions may struggle to respond if job structures shift all at once.

AI Is Eroding Job Autonomy Long Before Mass Layoffs Begin

For Amodei, the challenge has expanded beyond engineering into a crisis of coordination. He argued that governments should focus most of their attention on managing the transition. While risks tied to misuse and geopolitical tension remain manageable, the margin for error is narrowing.

Key policy pressures emerging from the debate include:

• Labor shifts occurring faster than existing retraining systems can handle.
• Regulatory gaps surrounding powerful general-purpose models.
• Rising inequality driven by automation of skilled work.
• Concentration of AI capabilities among a small number of major players.
• Limited global coordination on safety standards.

Some labor analysts believe disruption may arrive through job restructuring rather than outright replacement. Bob Hutchins, CEO of Human Voice Media, said professional roles are being broken into smaller, more closely monitored tasks. Algorithms increasingly manage workflows that were once controlled by individual workers.

According to Hutchins, this shift changes how work feels and functions. Creative and technical roles move from decision-making positions to verification roles. Workers check outputs rather than shape projects. Over time, this process can strip jobs of autonomy and reduce wages, even if titles remain unchanged.

Rather than asking whether machines will replace people, Hutchins argued that attention should shift to how work quality is altered. As tasks fragment and oversight increases, professional identity itself may erode. Governments and employers now face a challenge that extends beyond preserving employment to preserving meaningful work as AI capabilities continue to expand.