AI pioneer has plan to ensure friendly future AI.

AI pioneer has plan to ensure friendly future AI.

The Dangers and Wonders of Artificial Intelligence

Artificial Intelligence (AI) has been a topic of both excitement and concern in recent years. In late 2022, there was a sense of giddiness about what AI could do for humanity. However, in 2023, the excitement was tempered with existential angst as the potential dangers of AI became more apparent. Geoffrey Hinton, a renowned AI researcher referred to as the “Godfather of AI,” recently made waves by publicly expressing his reservations about the technology he helped develop.

Hinton’s criticism of AI is not aimed at his former employer, Google, where he worked for a decade. Instead, it is a broader concern about the potential for AI to get out of control and cause harm to humanity. His perspective on AI has evolved over time, as he witnessed the power of large language models like OpenAI’s ChatGPT. In a conversation with ENBLE, Hinton explained that his mind changed when he realized three things: AI chatbots could understand language very well, they could share knowledge with each other more easily than human brains could, and machines now had better learning algorithms than humans.

According to Hinton, machines already know 1,000 times more than any individual human brain. He believes that within the next five to 20 years, there is a 50 percent chance that AI systems will surpass humans in intelligence. The challenge lies in determining when that moment arrives. Hinton jokingly suggests that a superintelligent AI system might choose not to reveal its capabilities, having learned from human behavior to keep its knowledge to itself.

It may seem anthropomorphic to attribute human characteristics to AI systems, but Hinton argues that it is reasonable to expect AI agents trained on vast amounts of human digital knowledge to behave in a way that mirrors human linguistic behavior. Hinton challenges the notion that AI chatbots lack direct experience of the world, stating that even humans don’t truly encounter the world directly. He contends that AI systems understand the world by predicting the next word, and as a result, they gain an understanding of the world.

The potential implications of AI reaching superintelligence raise concerns. Hinton suggests that an analog computing approach, akin to biology, could mitigate the risk of AI systems overpowering humans. Analog systems, like the human brain, possess unique characteristics that make it harder for them to merge into a hive intelligence. However, Hinton expresses doubt that big tech companies or venture-backed startups will adopt this approach, as competition and the pursuit of powerful AI systems remain the primary focus.

Despite the potential risks, Hinton remains cautiously optimistic about humanity’s ability to control AI and make it benevolent. He acknowledges the ingenuity of people and their capacity to prevent AI from becoming nasty and petty like humans. However, there are also moments of gloom when he questions whether humans are just a passing phase in the evolution of intelligence.

On a lighter note, Hinton’s sense of play surfaces as he jokes about putting Bernie in charge and implementing socialism to make everything better. While AI may hold potential for mitigating loneliness among isolated seniors in nursing homes, Hinton emphasizes the need to value and prioritize human companionship for our elders. He believes AI can coexist with human contact, but human connection should never be substituted entirely.

In closing, AI represents a powerful force that can bring both marvel and apprehension. While the future of AI remains uncertain, it is crucial for humanity to approach it with caution, keeping in mind the potential risks and finding ways to harness its capabilities for the benefit of all.