I have to say, I don't know how to place AGI and ASI in the grand scheme of things. There's a sociological puzzle: do the people running AI (the Altman's of the world) actually believe AGI is around the corner? Are they making business decisions based on this impending event? What about military and industrial planners? Is this for real?

The Curve conference had much to say about these questions.

At the Curve conference in Berkeley, a rare gathering of AI insiders, the author found themselves immersed in “the room where AI happens”—a space buzzing with intense debates, groundbreaking ideas, and the presence of leading figures like Yoshua Bengio and Ben Buchanan. The event revealed the complex, multifaceted nature of AI’s future, especially through a heated debate between two camps: the “AI 2027” believers, who foresee a rapid, recursive leap in AI capabilities once research is automated, and the “AI as normal technology” advocates, who expect gradual progress constrained by real-world bottlenecks like regulation and infrastructure. This tension highlighted how difficult it is to predict AI’s trajectory, with some experts even proposing a third view—“jaggedness”—where AI excels dramatically in narrow tasks but remains flawed in others, challenging assumptions of uniform growth.

China’s role in AI emerged as a subtle yet revealing theme. Technical experts focused on China’s open-source AI projects, which, while not the most advanced, are widely adopted and rapidly improving, much like Chinese electric vehicles a decade ago. In contrast, policy discussions centered on geopolitics, export controls, and national security, creating a divide where the technical reality and geopolitical narratives rarely intersected. This split underscored differing mental models about what “winning” in AI means.

The conference also brought forward profound reflections on AI and creativity. Ted Chiang argued that AI-generated art lacks human intention and the capacity to astonish, as it recombines existing works without original thought. In contrast, Kevin Kelly embraced AI as a tool to shape creative direction, even offering his books to train AI, while Ken Liu likened AI’s creative evolution to early cinema’s journey toward complex storytelling. These perspectives, though divergent, collectively enrich the conversation about AI’s impact on art.