Tech • IA • Crypto
When Sanders and Trump agree, the situation is serious!
9.2/10TL;DR
A new roadmap from Google DeepMind outlines a path to superintelligence within a decade, as political tensions rise globally over control, ownership, and regulation of advanced AI.
KEY POINTS
DeepMind sketches path to superintelligence
A 60-page research paper led by Shane Legg, a co-founder of DeepMind and pioneer of artificial general intelligence (AGI), lays out a technical roadmap toward artificial superintelligence (ASI). AGI is defined as systems matching human performance across most cognitive domains, while ASI would exceed humans by orders of magnitude, comparable to thousands of coordinated experts working for years. The paper suggests this leap could occur within a decade, driven by accelerating progress.
Four technical routes identified
DeepMind highlights four main approaches: scaling compute power through more GPU infrastructure, coordinating multi-agent systems to create collective intelligence, advancing new model architectures beyond current large language models, and enabling systems to self-improve recursively. These pathways reflect a shift from single models toward complex, autonomous ecosystems of AI systems.
Debate over limits of language models
The roadmap enters an ongoing debate over whether large language models (LLMs) alone can reach AGI. While critics such as Yann LeCun long argued LLMs were insufficient, recent advances have narrowed that skepticism. Hybrid approaches combining LLMs with “world models”—systems that simulate physical reality—are gaining traction as a more viable route to higher intelligence.
Data and robotics become critical bottlenecks
Future progress increasingly depends on access to real-world data, not just internet text. Autonomous vehicles, robotics, and sensor-rich systems are expected to generate massive datasets needed to train next-generation AI. Countries leading in robotics, notably China, are rapidly expanding capabilities, while Europe lags with limited industrial scale.
Political tensions escalate globally
AI development is triggering sharp political reactions. In the United States, emergency restrictions have reportedly been applied to advanced models, while leaders across the spectrum—including Bernie Sanders and Donald Trump—are converging on the idea that the public should hold a stake in AI systems. This rare alignment signals growing concern over concentration of power.
Europe confronts strategic dependence
European policymakers are increasingly aware of reliance on foreign AI infrastructure and platforms. Limited investment—measured in hundreds of millions of euros versus billions per day in U.S. private spending—has left the region struggling to compete. The gap is compounded by the absence of major tech giants capable of funding large-scale AI research.
Corporate dominance fueled by private capital
Most AI breakthroughs are financed by cash flows from major tech firms such as Google, Amazon, and Microsoft. Their dominance allows sustained investment in compute, talent, and infrastructure, reinforcing a cycle that smaller economies cannot بسهولة replicate.
Social backlash and workplace resistance
Adoption of AI is uneven and often contentious. Studies suggest only 5–6% of companies report clear productivity gains so far, while up to 44% of younger workers actively resist or sabotage AI tools due to job security fears. This highlights a growing disconnect between technological capability and real-world integration.
Rising calls for redistribution mechanisms
Proposals are emerging to address AI-driven inequality, including sovereign wealth funds, public ownership stakes, and even universal income models. Some suggest citizens should benefit from AI similarly to how resource-rich nations distribute oil revenues, reflecting concerns about wealth concentration.
Regulatory dilemmas intensify
Governments face complex trade-offs: taxing AI could slow innovation in critical fields like healthcare, while failing to regulate risks widening inequality and monopolies. At the same time, unilateral regulation is difficult in a global մրց environment where competitors such as China continue rapid development.
CONCLUSION
The race toward superintelligence is accelerating faster than political and economic systems can adapt, raising urgent questions about governance, ownership, and societal impact that remain unresolved.