Google DeepMind Institute published an essay, "What Comes After AGI?", arguing that AGI is a waypoint on a broader intelligence continuum rather than an endpoint. It lays out four potential routes toward artificial superintelligence spanning capability boundaries, the industrialization of cognition, technical pathways and recursive self-improvement, and stresses that measurement, forecasting and governance matter as much as raw capability.
AGI
通用人工智能
AI matching or surpassing humans across a broad range of tasks, with contested definitions and paths to realization.
Curated Intelligence
Even before AGI is achieved, frontier research has begun systematically examining what comes after it, framing general intelligence as a continuum and mapping possible paths from AGI to artificial general superintelligence (ASI). The work stresses the need for rigorous, reliable ways to track progress, given a shift that could prove far more profound and rapid than AGI itself.
In an interview with The Economist, Elon Musk predicted that AI will surpass the combined intelligence of all humanity around 2031, painting an optimistic picture of an AI-and-robot-driven age of abundance while conceding the risk of losing control is non-zero. He reiterated his July proposal for AI firms to grant rivals one to two weeks of early model access for safety and misuse testing, with government intervention as a backstop.
Microsoft CEO Satya Nadella said at a roundtable that he has become "AGI-pilled," arguing that the industry must think about what political-economic structures AGI will produce. He nonetheless denied that frontier models such as Claude are conscious, aligning with Mustafa Suleyman's view that companies should not grant AI models rights or moral status, and stressed that broad economic diffusion of AI is the industry's most pressing issue.
Microsoft CEO Satya Nadella said at a post-Ignite roundtable that he is "AGI-pilled" regarding how AGI could reshape the political economy, while rejecting the idea that frontier models like Claude are conscious. He echoed Mustafa Suleyman's opposition to granting models moral status and argued that a world needing more than one or two firms requires as many models as firms, with sovereign data preventing centralized control.
Prominent AI observer Ethan Mollick tested Anthropic's new Claude Fable 5.2 ahead of public launch and found it highly impressive, building on the already strong Opus 5.5. He frames AGI as matching most human experts on most computer-based intellectual tasks and asks whether this model and imminent OpenAI and Google rivals now cross that limited threshold.
Even as GPT-6 Astra's launch prompted Jensen Huang to declare that AGI has arrived, co-creators of today's models are leaving OpenAI, Google DeepMind and Meta to found new labs that raise billions with no product. The piece sorts 45 such neolabs by research bet, maps the limitations of current AI they aim to overcome, and argues they are wagering that the existing recipe alone will not be enough.
Framing science fiction as descriptive rather than predictive, the essay maps a decade of real AI progress against classic stories such as The Lifecycle of Software Objects and Galatea 2.2, scoring each on how closely it anticipated the Transformer-driven rollout of AI to billions of users. The author argues that sci-fi's value lies not in forecasting technical roadmaps but in showing the concrete mechanisms by which change unfolds.
Speaking at China's World Internet Conference, DeepSeek engineer Gao Xi said the company will keep releasing model weights despite higher safety risks, aiming to compress a ten-year wait for AGI into one year, and one year into "tomorrow." It is a strategic stance rather than a concrete release, with no model details, timeline, or safety measures disclosed, so the substantive information gain is limited.
Michael I. Jordan, a leading machine learning pioneer and Inria chair holder, harshly criticized current AI doom narratives as a 'grand moralizing performance,' dismissing superintelligence as pure science fiction. He urged the industry to stop chasing AGI hype and instead leverage open-weight models integrated with local domain knowledge to solve specific business problems cost-effectively without massive data centers.