Google DeepMind CEO Demis Hassabis sat down with 60 Minutes to discuss the future of Artificial Intelligence (AI). The interview touched on Gemini, a chatbot platform currently available to trusted testers. Notably, Gemini has features like contextual memory, allowing it to remember key details from previous conversations and personalize its responses.
In terms of capabilities, Gemini Live is expected to build upon this feature set, with the ability to recall information from past conversations for improved context. Additionally, a “10-minute memory” of the current conversation will also be added.
The interview also delved into Google DeepMind’s plans for AI beyond just revealing information. Hassabis stated that the goal is for AI to interact with humans in everyday life, such as booking tickets and shopping online. This aligns with Sundar Pichai’s previous mention of Project Mariner, which aims to bring this functionality to Gemini.
Regarding AGI (Artificial General Intelligence), Hassabis estimates it may be 5-10 years away. In 2030, he envisions a system that can deeply understand its surroundings and be embedded in daily life. However, when asked about the possibility of self-awareness, Hassabis expressed skepticism. He noted that while systems may exhibit behaviors similar to conscious beings, true consciousness remains uncertain.
The interview highlighted two key factors in determining whether machines are conscious: 1) exhibiting behavior similar to a conscious being and 2) sharing the same substrate (carbon matter). Despite advancements in AI, it’s unclear if machines can replicate human consciousness.
Source: https://9to5google.com/2025/04/21/demis-hassabis-gemini-next