Gemini Will Understand Feelings? Google's New Technology Makes It Merinding!

Google DeepMind CEO, Demis Hassabis (photo: x @ATTSquared)

JAKARTA - In an exclusive interview with program 60 Minutes, Google DeepMind CEO, Demis Hassabis, discussed the future of artificial intelligence (AI). This includes the latest developments in the Gemini project, the possible presence of Artificial General Intelligence (AGI), to the issue of AI's self-awareness.

This interview features more footage from the Project Astra app, which is currently only available to trusted testers. One of the interesting things is Astra's personal greeting: "Hello, Scott. It's nice to see you again."

This feature is not yet available on Gemini Live, which currently has an internal memory system that remembers important information from previous conversations to provide a better context and personalization. There is also a "10-minute memory" feature of active conversations. These features are expected to be integrated into Gemini Live in the near future.

Not only that, but Astra also showed a brief demonstration that was running on smart glasses.

Hassabis revealed that DeepMind is training Gemini's AI model not only to understand the world, but can also act in it, such as booking tickets and shopping online. This feature sounds like part of Project Mariner, previously mentioned by Google CEO Sundar Pichai, and will be coming soon at Gemini this year.

AGI Predictions And Potential Self-Awareness

Regarding AGI, Hassabis still believes that the form of general artificial intelligence will materialize in the next 5'10 years. When asked by 60 Minutes about what AGI will look like in 2030, he replied:

...we will have a system that truly understands everything around us in a very deep and nuanced way and is part of our daily lives.

Regarding the possibility of AI becoming self-aware, Hassabis replied that currently there is no system that shows any self-awareness or form of awareness.

"I don't think any current system feels self-aware for me... But theoretically, it might happen," he said.

When asked if self-awareness was an explicit goal of DeepMind, Hasssabis emphasized:

Not explicitly. But it could happen implicitly. These systems may develop a kind of feeling of self-awareness. That might happen.

He added that understanding oneself and others is the basis of something similar to awareness. However, he stressed that although AI can mimic conscious human behavior, the fundamental difference in substrate' between humans (carbons and biological brains) and machines (silicon) is a philosophical and technical challenge about the true meaning of machine awareness.

With this rapid development, Google DeepMind appears to be at the forefront of shaping the future of AI that is not only smart, but also has the potential to become an integral part of human daily life with or without self-awareness.