Timothy Garton Ash, a prominent political commentator, reports from Silicon Valley that artificial intelligence (AI) appears to be advancing faster than humans can effectively control. According to his account published in The Guardian, experts there expect AI’s progress to accelerate dramatically over the next few years.
What experts are warning about
The experts Garton Ash spoke with say we are close to a threshold where human-created knowledge can begin to improve itself and be refined across successive models. That self-improvement could start an exponential trajectory, and some systems may become, in several respects, more intelligent than their creators.
Diverging outlooks: benefits versus risks
The report cites Elon Musk, who foresees enormous material benefits from AI but also assigns a 10–20% chance that weaponized or “killer” robots could pose deadly risks to humanity. Geoffrey Hinton, often described as one of the “fathers” of modern AI, is reported to consider even darker, ‘hellish’ scenarios to be plausible.
Why this matters
These observations matter because rapidly accelerating, partially self-improving AI systems leave less time for governance, oversight, and mitigation than previously assumed. The range of expert opinion — from optimistic forecasts of wealth creation to warnings of catastrophic outcomes — highlights the urgent need for policymakers, researchers, and the public to engage seriously with the social, ethical, and safety implications.
Conclusion
Garton Ash’s account suggests that the accelerating pace of AI development is no longer solely a technical issue: it raises immediate policy and safety challenges, with experts sharply divided over likely outcomes.
