Alex Hern
Showing 1–8 of 8 transcripts.
- The Economist9 min
The dangers of accidentally creating a conscious AI | The Economist
Researchers are shifting toward mechanistic interpretability to probe the internal architecture of Large Language Models, revealing a "global workspace" mechanism in Claude Sonnet that parallels human consciousness theories. While AI labs currently view this as a scientific inquiry rather than an engineering goal, philosophers warn that the accidental emergence of sentient entities capable of suffering poses a moral catastrophe. Consequently, experts recommend pausing frontier development to establish an intentional ethical framework before such capabilities are realized.
- The Economist7 min
Why the AI race won't have a winner | The Economist
Europe is pursuing a fragmented AI strategy that leverages sovereign capabilities in specific domains like audio processing and symbolic reasoning to capture a projected $1 trillion market without relying on a single US provider. National governments in France, Luxembourg, and other states are accelerating industrial policy to secure strategic power comparable to oil, prioritizing the ownership of AI infrastructure to enhance civil servant productivity. This approach aims to stabilize the region's economy against global commercial imbalances while addressing demographic challenges by using AI to augment, rather than replace, the aging workforce.
- The Economist9 min
AI has got better at hacking—how big a risk is it?
Anthropic withheld its latest model, Mythos, from public release after it demonstrated the ability to autonomously identify and exploit critical software vulnerabilities, such as a 27-year-old OpenBSD flaw, with minimal human intervention. This decision follows evidence that AI scaling laws now enable automated systems to accelerate the vulnerability exploitation cycle from years to mere hours, drastically outpacing current defensive capabilities. While Anthropic employed responsible disclosure by patching the specific bug before sharing details, the incident highlights a growing risk that open-source models will soon democratize high-level cyberattacks.
- The Economist6 min
Google DeepMind’s boss on AI, power, God and what’s next | The Economist
An AI researcher outlines a vision where artificial general intelligence serves as a fundamental scientific instrument to solve existential challenges like disease and energy crises, rather than a tool for corporate dominance or "making God." Despite a fragmented geopolitical landscape that hinders the preferred international cooperation model, the speaker advocates for cautious optimism and the establishment of minimum safety standards agreed upon by leading labs. This approach aims to ensure the safe delivery of AGI within roughly five years, fulfilling a lifelong mission that prioritizes global benefit over personal accolades.
- The Economist8 min
Why does Elon Musk want to put data centres in space? | The Economist
Elon Musk, Tom Standage, Alex Hern
Elon Musk has merged XAI into SpaceX and filed an FCC application to deploy one million satellites, aiming to establish orbital data centers powered by sun-synchronous orbits to achieve terawatt-scale annual capacity. While Musk targets lunar manufacturing and mass drivers to lower launch costs, OpenAI's Sam Altman and industry analysts currently dismiss the concept as technically infeasible due to prohibitive expenses and unrepairable hardware failure rates. Despite this skepticism, competitors including StarCloud, Google, Jeff Bezos, and Eric Schmidt are advancing their own pilot projects or securing launch capacity to test the viability of space-based computing.
- The Economist8 min
Could the AI bubble pop?
Market analysts describe a speculative frenzy where investors treat AI startups as high-stakes lottery tickets, driven by the potential for a single entity to capture a $1.46 quadrillion monopoly by achieving superintelligence. Unlike historical infrastructure booms that left behind tangible physical assets, this sector faces unique risks from rapidly obsolete specialized hardware and the disruptive proliferation of free open-source models that could strip commercial value from paid solutions. Consequently, the ultimate economic outcome remains uncertain as the market grapples with whether the industry will result in a winner-takes-all dominance or a fragmented collapse similar to the 2001 telecom crash.
- The Economist8 min
Why are cybercriminals richer than ever?
Criminal gangs have evolved from data theft to aggressive "wrecking" attacks that destroy systems to halt operations, with young British hackers recently collaborating with Russian groups to target major entities like Jaguar Land Rover. The JLR incident alone caused an estimated £1.5 billion in damage, exceeding the total global ransom income of $1 billion in 2024, while exposing critical vulnerabilities in non-core infrastructure and cost-optimized outsourcing models. In response to this economic devastation, the UK government is considering a legal ban on ransom payments to disrupt the criminal market, alongside stricter transparency mandates that would require immediate reporting of severe cyber disruptions.
- The Economist11 min
Will AI take Apple’s crown?
Apple's WWDC showcased a major "liquid glass" visual overhaul for iOS while exposing strategic vulnerabilities in its "Apple Intelligence" rollout, which has missed promised feature deadlines and lags behind competitors like Google. The company also faces significant legal and operational challenges, including a court injunction restricting developer communication that threatens its App Store revenue and a supply chain heavily dependent on China. Despite high consumer retention in the premium market, these factors combine with a shifting AI landscape to create uncertainty for Tim Cook's leadership and the long-term valuation of Apple's hardware ecosystem.