From AGI to ASI; the AI’s unprecedented growth

In recent AI developments, we’ve seen Geoffrey Hinton, often hailed as one of the “Godfathers of AI,” speak up since his retirement from Google earlier this year. Hinton’s claim to fame is the “backpropagation” algorithm, a key element in the success of deep learning technologies, which are the backbone of today’s generative AI models. For his groundbreaking work, he’s been awarded the Turing Award, often referred to as the computer science equivalent of the Nobel Prize.

But here’s where it gets interesting: Hinton used to believe that AI would take a good 50 to 60 years to surpass human intelligence. However, now he’s got a much more urgent perspective, suggesting it could happen in as little as five years. Why the hurry? Well, he’s looking at the rapid strides in generative AI, particularly through large language models (LLMs).

His thoughts are kind of aligned with those of another AI enthusiast, Ray Kurzweil. Kurzweil has predicted that by 2029, computers will match human intelligence and even take it further with what he calls the “Singularity,” a point where AI multiplies our effective intelligence immensely.

In an interview, Hinton went as far as saying that present top AI models, including those from OpenAI and Google, already show real intelligence and reasoning abilities. He believes that these models could eventually develop consciousness, although they’re not quite there yet.

Next 5 years crucial

Hinton thinks that within the next five years, advanced AI models might outdo humans in reasoning. He even believes that humans could become the second smartest species on the planet in the near future. But he’s quick to emphasise that there’s a lot of uncertainty about what’s coming next.

The pace of AI growth is putting pressure on global governance. The EU is in the final stages of debating comprehensive AI legislation, known as the AI Act. But the US is worried that this law could favor the big players and hinder innovation.

Dealing with various countries’ AI regulations could be a major headache, especially for smaller businesses trying to navigate through the maze. The G7 leaders are expected to establish international AI regulations, but the absence of China and many EU countries raises questions about the effectiveness of these agreements.

As AI accelerates at an unprecedented rate, it’s absolutely critical for governments, businesses, and society to come together for comprehensive global AI governance. The ethical and responsible use of AI is of paramount importance, given the profound changes AI could bring.

From AGI to ASI

AI’s journey from where it is now to artificial general intelligence (AGI) and maybe even artificial superintelligence (ASI) could be a game-changer for humanity. AGI represents AI’s ability to excel in a range of cognitive tasks, while ASI goes beyond the collective intelligence of all humans. An important step in this journey is “recursive self-improvement,” where AI systems enhance their code.

 The ethical and technical questions surrounding ASI are complex and unpredictable. After all, we’ve never created a superintelligent AI before. Human intervention might slow down the advancement of AI and ensure that we have the right safeguards in place to handle the risks.

In the end, AI’s rapid evolution calls for thoughtful global governance and ethical considerations. Our future as a species could very well hinge on how we address the challenges and use AI wisely and responsibly. Collaboration and unity are essential as we navigate this transformative AI era.