Artificial intelligence : are we ready for the day after?
At the World Economic Forum Annual Meeting 2026, the heads of Google DeepMind and Anthropic discussed the imminent arrival of Artificial General Intelligence and its societal ramifications. The conversation shifted from technical milestones to the urgent need for a “battle plan” to navigate the risks of labor displacement, biosecurity, and geopolitical instability. While the potential for scientific breakthroughs is unprecedented, the experts warn that our window for establishing global safety governance is rapidly closing.
Points clés
- Dario Amodei (Anthropic) and Demis Hassabis (Google DeepMind) reunited for a historic sequel to their 2025 Paris debate.
- Amodei predicts AI models capable of Nobel-level performance across multiple fields could emerge by 2026 or 2027.
- Anthropic’s revenue has scaled exponentially from $100 million in 2023 to a projected $10 billion in 2025.
- Hassabis remains more cautious, suggesting a 50% chance of human-level cognitive capabilities by the end of the decade.
- Both leaders identified “closing the loop”—AI systems autonomously researching and building better AI—as the primary driver of development speed.
- Amodei expressed grave concerns regarding “technological adolescence” and the risks of autonomous systems being misused for bioterrorism.
- Google DeepMind is focusing on “AlphaFold-like” breakthroughs to solve diseases and energy needs as a way to demonstrate AI’s “unequivocal good.”
- The panel criticized current US policy on selling high-end chips to geopolitical adversaries, comparing it to selling nuclear secrets for profit.
- Amodei reiterated a previous warning that 50% of entry-level white-collar jobs could be disrupted within a 1-to-5-year timeframe.
- Both experts emphasized that while technical risks are solvable with time, global cooperation between the US and China is currently insufficient.
À retenir
If you were hoping for a slow transition into the post-work era, I have some bad news: the people building the future are currently arguing over whether we have two years or five until the machines outsmart us. It’s comforting to know that while your job might be automated by next Tuesday, at least AI will be busy curing cancer—assuming it doesn’t accidentally help a bad actor reinvent the plague first. My recommendation? Start practicing your “meaningful hobbies” now, because according to these gurus, we’re either heading for the stars or becoming the most sophisticated paperweights in the galaxy.
Sources
Quiz sur la vidéo: 5 questions
