From AI to AGI: The Quest for True Intelligence - Oriol Vinyals at Google Gemini Conference
An inspiring conference by Oriol Vinyals, technical leader of Gemini at DeepMind, where he explored the latest advancements in Google Gemini and the future role of AI in achieving Artificial General Intelligence. The talk covered groundbreaking possibilities and the path toward truly intelligent systems.
Oriol Vinyals at UPC Barcelona
In January 2025 I attended a talk at UPC in Barcelona by Oriol Vinyals, one of the technical leads of Gemini at Google DeepMind and one of the most influential researchers in modern AI. The title set the tone: From AI to AGI, the quest for true intelligence.
Inside Google Gemini
The first part of the talk covered the latest advances in Gemini, Google's family of multimodal models. Seeing how a single model can work with text, images, audio and code together makes it clear how quickly the field has moved from narrow, single-task systems to general-purpose models.
Hearing it from someone who builds these systems gave a much more grounded view than the usual headlines: the progress is real, and so are the engineering challenges behind it.
The path from AI to AGI
The second part looked ahead to Artificial General Intelligence: what it would mean for AI systems to reason and learn across any domain, and what still separates today's models from that goal.
Whatever your view on timelines, the direction is clear. Models are becoming more capable, more general and better at acting, not just answering.
What frontier models change for everyday companies
Talks like this are usually about the frontier, but the practical consequence reaches every business. The capabilities that were research a couple of years ago, such as understanding documents, extracting data or reasoning over a process, are now available through an API.
For companies, the question is no longer whether AI can do a task, but how to integrate it into real workflows reliably. That is where most of the value, and most of the work, lies.