In Vancouver, at the vast convention center, a major breakthrough in artificial intelligence could be unfolding, as over 16,000 computer scientists gathered for NeurIPS, the Conference on Neural Information Processing Systems, the largest annual AI event. This gathering, which began in 1987, has grown from a small, niche affair to a critical platform where AI's top minds converge, shaping the future of technology and the global economy.

AI experts like Ilya Sutskever, former chief scientist at OpenAI, and Fei-Fei Li from Stanford shared their visions of AI's future. Sutskever noted that as AI's reasoning capabilities expand, its unpredictability also grows. Li, in turn, discussed the importance of spatial intelligence, arguing that relying solely on 2D data was like building AI for a "flat earth."

The conference, once a small gathering, now serves as a high-profile venue for corporations to promote their technologies. Major companies such as Alphabet, Meta, and Microsoft used the event to make key AI announcements. The crowds were so large that the event started a day later than usual to accommodate all attendees, with overflow from a Taylor Swift concert creating hotel booking challenges.

Longtime attendees, including AI pioneer Yann LeCun, reflected on how much the event had changed over the years. What used to be an intimate academic gathering is now dominated by corporate influence, with venture capitalists and investors playing a prominent role. The number of conference sponsors has grown significantly, from nine in 2006 to over 120 this year.

Research at the conference has also exploded, with ten times as many papers accepted this year compared to a decade ago. One emerging trend is the growing focus on AI evaluations and measurement science, highlighted by a paper on AI models predicting images at higher resolutions. Meanwhile, a 10-year-old, Harini Shravan from India, made history as the youngest person ever to have a paper accepted, using AI to transform a 3,000-year-old story into a musical.

At a side event, Jeff Dean from Google DeepMind shared his perspective on AI's future, advocating for more modular, sparse, and complex AI models—what he referred to as his "spicy" view.

Disclaimer: This image is taken from Reuters