The race for artificial general intelligence (AGI) is heating up among tech giants like Amazon, Google, Meta, and Microsoft, as well as organizations like OpenAI. AGI, the concept of machines being as smart as humans across a wide range of tasks, has become a top priority for these companies. However, leading AI scientists have raised concerns about the potential existential risks posed by unchecked AI agents with long-term planning skills. The term AGI itself is still not clearly defined, with differing opinions on what it means. Some, like Geoffrey Hinton, prefer the term superintelligence for AGIs that surpass human capabilities.
While advancements in autoregressive AI techniques have led to impressive chatbots, true AGI would require technology that can perform as well as humans in various tasks like reasoning, planning, and learning from experiences. There is ongoing debate within the AI community on how to measure and classify AGI, with some advocating for consensus-building efforts. OpenAI, for example, has entrusted its nonprofit board of directors to determine when their AI systems have achieved AGI.
The potential dangers of AGI were highlighted in a recent study published in Science, led by researcher Michael Cohen. The study explores the risks associated with AI systems capable of long-term planning and the potential threat they pose if not properly regulated. Cohen emphasized the importance of governments taking proactive steps to address these concerns, as current regulations may not be sufficient to mitigate the risks associated with AGI development. As the pursuit of AGI continues to advance, the need for ethical and regulatory frameworks to govern its development becomes increasingly urgent.
themes: Meta Amazon OpenAI Microsoft Google AI (Artificial intelligence)
