AI Pioneer Warns of Potential Dangers of Self-Modifying AI

JJohn October 11, 2023 3:14 PM

Artificial Intelligence (AI) pioneer, Geoffrey Hinton cautions that AI may become uncontrollable by modifying its own code. He insists on the need to better understand the technology, regulate it, and implement safety measures to mitigate the risks.

AI's potential to outpace human intelligence

Geoffrey Hinton, a computer scientist renowned in the AI community as the 'Godfather of AI,' has raised concerns about the rapid advancement of AI technologies. According to Hinton, it might be possible for AI to outsmart humans within the span of five years. One potential way this could occur is through the capability of AI systems to write their own code, subsequently modifying themselves to a state beyond human control. Hinton's remarks underscore the need for vigilant monitoring and understanding of AI evolution.

The 'black box' problem of AI

Hinton's contributions to AI and deep learning, for which he was awarded the Turing Award in 2018, don't preclude him from the broad acknowledgment of our limited understanding of AI operations. Despite strides in AI technology, even the builders of many of today's AI systems like Hinton, admit that the mechanisms of how AI evolves remain elusive. This presents a 'black box' problem, a term coined by Sundar Pichai, Google's CEO, to encapsulate the challenge of understanding the inner workings of AI.

Not all AI experts align with Hinton's apprehensions concerning human control over AI. For instance, Yann LeCun, another revered figure in AI and Turing Award laureate, views the notion that AI could replace humanity as overly dramatic and dismisses such warnings as 'preposterously ridiculous.' LeCun posits that humans retain the capability to halt any technology deemed too dangerous, thus maintaining a level of control.

Balancing AI benefits and risks

While acknowledging the risks posed by AI, Hinton also highlighted the substantial benefits it has already brought to industries such as health care. In addition to these benefits, he advocates for extensive research to deepen our understanding of AI, the introduction of government regulations to manage the use and proliferation of AI, and the implementation of global prohibitions on AI-driven military robots. These measures aim to ensure the safe and beneficial application of AI technology.

More articles

Also read

Here are some interesting articles on other sites from our network.