The recent executive order signed by Donald Trump, advocating for the term 'super intelligence' instead of 'artificial intelligence,' has brought renewed attention to the nuanced distinctions between these concepts. Trump argued that 'artificial' carries an element of fakeness, suggesting 'super intelligence' better captures the advanced capabilities of evolving AI technologies. While the political push for new terminology is notable, understanding the fundamental differences is crucial.
What is Superintelligence?
The concept of superintelligence predates recent political discourse. Hungarian-American mathematician John von Neumann first explored the possibilities of machines surpassing human intellectual capabilities in the early 1950s. Later, in 1965, British statistician IJ Good introduced the idea of an 'ultraintelligent machine' capable of designing even more intelligent machines.
The modern definition, widely popularized by Nick Bostrom in his 2014 book Superintelligence: Paths, Dangers, Strategies, defines superintelligence as an intellect that 'greatly exceeds the cognitive performance of humans across virtually all domains.' This implies not just superior processing power in specific tasks, but a comprehensive cognitive superiority across the entire spectrum of intellectual activities.
Superintelligence vs. Artificial Intelligence: The Core Distinction
The primary difference lies in scope and capability. Artificial Intelligence (AI) is a broad and overarching field encompassing any machine or system that can perform tasks typically requiring human intelligence. This includes everything from simple algorithms to complex neural networks designed for specific functions, often referred to as 'narrow AI.'
Superintelligence, however, is a theoretical and hypothetical stage of AI development. It represents a level of intelligence that is not only 'good at its job' but is 'far better' than humans across a wide range of intellectual activities simultaneously. For instance, while an AI might excel at discovering a drug faster than human researchers, a superintelligence would demonstrate superior capabilities in drug discovery, mathematics, programming, language comprehension, research, and general cognitive tasks all at once.
The Stages of AI Evolution
While not a standardized technical benchmark, a simplified framework often describes AI's progression in four stages:
- Narrow AI: Designed for specific tasks (e.g., image recognition, chess playing).
- General AI (AGI): Hypothetical AI capable of understanding, learning, and applying intelligence across a wide range of intellectual tasks, similar to a human.
- Human-Level General Intelligence: AGI that matches human cognitive performance.
- Superintelligence: AI that significantly surpasses human intellectual abilities across virtually all domains.
Trump's Terminology Shift and Expert Reception
Donald Trump's executive order aims to redefine the terminology used by the administration, hoping to inspire a broader adoption of 'super intelligence.' He suggested this change during a speech at the UN General Assembly. However, experts in the field remain cautious about the widespread acceptance of this new term at the current stage, noting that the term 'superintelligence' itself is not new, but its political endorsement as a replacement for 'artificial intelligence' is.
It remains to be seen whether other nations or the broader scientific community will follow suit in adapting this terminology, or if it will primarily remain a political designation within a specific administration.