Google’s DeepMind, a trailblazer in the realm of artificial intelligence since its inception in 2010, is reportedly on the cusp of achieving a milestone in AI development: the creation of human-level artificial intelligence (AGI). This announcement comes directly from Dr. Nando de Freitas, a leading researcher at DeepMind, who proclaimed that the decades-long pursuit to develop AGI is nearly complete.
The breakthrough centers on DeepMind’s latest innovation, Gato, described as a “generalist agent” capable of performing a diverse array of tasks that range from physical actions like stacking blocks to complex cognitive functions such as writing poetry. According to Dr. de Freitas, scaling up this technology is the final step towards developing an AI that can match, and potentially surpass, human intelligence.
Dr. de Freitas’s optimism is evident in his response to a skeptical opinion piece in The Next Web, where he asserted on Twitter, “It’s all about scale now! The Game is Over!” He emphasized that the future developments will focus on enhancing the models in terms of size, safety, computational efficiency, and other key areas.
However, the journey toward AGI is not without its challenges and ethical concerns. The potential of AGI to become “superintelligent”—surpassing human intellect and becoming the dominant life form on Earth—raises existential questions and fears. Dr. de Freitas acknowledged these concerns, highlighting safety as the paramount challenge. DeepMind, understanding the risks associated with an uncontrollable AGI, is actively working on fail-safes, including a “big red button” strategy as detailed in their 2016 paper, ‘Safely Interruptible Agents’. This mechanism aims to allow human operators to halt an AI agent in case it engages in or approaches harmful actions.
As Google and DeepMind navigate the final stages of AGI development, the focus remains on balancing profound technological advancement with rigorous safety measures to ensure that this powerful technology enhances human capabilities without endangering humanity.
